--- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@...> wrote:
>
>
>
> Hi Mike,
>
> If you have a Position delta just add it to the Position with:
>
> ch0->Position += Delta;
>
> Position is a 64-bit floating point value (52 bits of integer precision).
>
>
> Configure the Axis as Input Mode of "User Input".
>
>
> BTW another approach that I usually find easier to implement in the FPGA is to just latch and send the bottom N bits of the absolute position.  KFLOP can then handle the "wraps" and keep track of the absolute position.  I can show you simple code to handle the wraps if you are interested (basically sum the changes).
>
> Regards
> TK
>
>
>
> ________________________________
> From: tesotronics <mike.bax@...>
> To: DynoMotion@yahoogroups.com
> Sent: Friday, June 7, 2013 4:19 AM
> Subject: [DynoMotion] Re: Philips 432 encoder adapter PCB for KFLOP
>
>
>
>  
> Hi TK,
>
> I got the frequency up to ~3 KHz, which is better but still a bit slow. I can further increase the frequency but that will go at the expence of resolution and noise.
>
> I've been looking into the C samples to try and understand how I can implement my bit-bang algorithm. Suppose I've "bit banged in" the delta position in counts since previous readout, and I have that number in a variable. What function can I then use to add the position increment/decrement to the old position in order to generate a position update? I could not find such functions in the example code.
>
> Regards,
>
> Mike
>
> --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote:
> >
> > Hi Mike,
> >
> > I suppose there is no way to increase the 1.5KHz.  That is a somewhat slow update rate.
> >
> > At a Max (Average) Speed of 1MHz that would represent sudden changes of 1MHz/1.5KHz ~ 700 counts.  So you would need ~ 11bits (10 bits plus sign).  You might send the position serially by doing something like SPI.  Even if it takes 100us for KFLOP to bit bang in the data is still much faster than the 1.5KHz update rate.
> >
> > Basically a 3 IO bit interface.  One KFLOP Output bit to latch 12-bits of position, One KFLOP output bit of clock to shift the 12 bits out.  One input data bits.  For multiple axes you could just add one additional input.
> >
> > What do you think?
> >
> > Regards
> > TK
> >
> >  
> >
> >
> >
> > ________________________________
> > From: tesotronics <mike.bax@>
> > To: DynoMotion@yahoogroups.com
> > Sent: Tuesday, May 21, 2013 1:30 PM
> > Subject: [DynoMotion] Re: Philips 432 encoder adapter PCB for KFLOP
> >
> >
> >
> >  
> > Hi Tom,
> >
> > I have been able to translate the legacy analog signal from the Philips 432 encoder to a step and direction signal.
> > However, the step signal is not varying in frequency as the speed changes like you would expect. Instead of that, the step signal gives bursts of 10 MHz pulses, occuring at an interval of around 1.5 KHz. The number of pulses in one burst represents the position change.
> > This is a bit of an odd way of measuring, but we inherit it since this is the way the decoding of the Philips encoder heads should be done.
> >
> > The question now rises what the best way of transfering the position to KFLOP would be.
> >
> > One possible scenario is that I convert the direction and (burst) step signals to quadrature A and B. But since the bursts are 10 MHz, translating the step signal will result in an A and B signal of 2.5 MHz, which is (probably) too high since you have spec'ed the inputs up to 1 MHz.
> >
> > I have started on an algorithm that will lower the A and B signal frequency to lower than 1 MHz, but this will logically introduce a small time lag. Introducing this small lag is likely not the most elegant way of doing it (transferring position I mean).
> >
> > Besides hooking up a parallel bus for each axis, do you possibly have another idea how we can do this?
> >
> > Regards,
> >
> > Mike
> >
> > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote:
> > >
> > > Hi Frank,
> > >
> > > TTL Encoder Input is probably the simplest.
> > >
> > >
> > > To connect to a KFLOP A B Encoder input you must output A B Quadrature not "pulses".  So I don't really understand your question.  KFLOP samples the A B signals at 16.67MHz.  There is a User programmable digital filter on each of the A B inputs that defaults to 7 (consecutive samples before a change will be accepted).
> > >
> > > You could possibly go 8-bits parallel and just have KFLOP read the position from your FPGA.  The main disadvantage would be the number of pins required. 
> > >
> > >
> > >
> > > In the Max Speed calculation there is a multiplication by 60 to change to units of minutes but then the answer is reported as in seconds.  0.1um resolution at 1MHz would be 6000 mm/minute.
> > >
> > > Regards
> > > TK
> > >
> > >
> > >
> > > ________________________________
> > > From: frank_19_88 <frank_19_88@>
> > > To: DynoMotion@yahoogroups.com
> > > Sent: Monday, May 20, 2013 11:47 PM
> > > Subject: [DynoMotion] Re: Philips 432 encoder adapter PCB for KFLOP
> > >
> > >
> > >
> > >  
> > > Hi TK,
> > >
> > > What is the fastest and most accurate way to communicate the axis position to the Kflop?
> > > - TTL Encoder Input's
> > > - ADC input
> > > - Or are there other ways
> > >
> > > Is it possible to detect 127 pluses every 90 us with a pulse length of 4e-7 (2,5 MHZ)?
> > > Because we don't have a square wave.
> > >
> > > Every pulse indicates 0,1 ÃÆ'‚µm which means:
> > > 0,1 ÃÆ'‚µm * 1 MHZ * 60 seconds = Max movement M/S
> > > 0,1 * 10 ^ -6 * 1 * 10 ^ 6 * 60 = Max 6 M/S
> > >
> > > With kind regards,
> > >
> > > Frank
> > >
> > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote:
> > > >
> > > > Hi,
> > > >
> > > > Regarding the KFLOP Encoder Input's max frequency:ÃÆ'‚  The bottle neck is an 8-bit register that the KFLOP DSP reads from the KFLOP FPGA every 90us.ÃÆ'‚  So the maximum rate is +/-127 counts/90us = 1.411e6 quadrature counts per second.ÃÆ'‚  We spec the inputs for 1MHz.
> > > >
> > > > KFLOP counts quadrature transitions (not complete cycles).
> > > >
> > > > KFLOP does not currently have any Step and Direction Inputs.
> > > >
> > > > Good luck on your project.
> > > >
> > > > Regards
> > > > TK
> > > >
> > > >
> > > >
> > > >
> > > > ________________________________
> > > > From: tesotronics <mike.bax@>
> > > > To: DynoMotion@yahoogroups.com
> > > > Sent: Monday, May 20, 2013 11:49 AM
> > > > Subject: [DynoMotion] Philips 432 encoder adapter PCB for KFLOP
> > > >
> > > >
> > > >
> > > > ÃÆ'‚ 
> > > > Hi,
> > > >
> > > > I'm working on an adapter PCB that enables the use of old Philips 432 encoders together with KFLOP. The PCB will give the user the possibility to convert a CNC mill with a Philips 432 control to a control with KFLOP without having to replace the original encoders that aren't supported by KFLOP.
> > > >
> > > > I'm working on the VHDL code at the moment and I have a few questions:
> > > >
> > > > -For the encoder inputs of KFLOP, what is the absolute maximum frequency of the A & B quadrature signals that the system can work with?
> > > >
> > > > -Does the KFLOP increment position on each edge of one period of the quadrature signal ( on all 4 edges, two rising and two falling)?
> > > >
> > > > -Can KFLOP handle a step and direction signal as input from an encoder system? If so, what would be the maximum frequency?
> > > >
> > > > I'm looking forward to finishing this product so conversions to KFLOP control for Philips 432 systems can become less costly and in a shorter about of time. I hope his forum can answer my questions so I can get the product to market quickly.
> > > >
> > >
> >
>