BCSB_motion

I've successfully set up a BISS-C encoder on an ACC-84E on Turbo hardware once, so take what I'm suggesting with a grain of salt. I'll break out the SerialEncCtrl and SerialEncCmd setups from the above registers to confirm that they're as intended: SerialEncCtrl: - Serial Frequency: 8.333_ Mhz - Trigger on Phase clock, falling edge - no trigger output delay - BiSS-C protocol SerialEncCmd: - CRC mask = $15 - BiSS-C protocol - (support of MCD bit=0)? - SerialEncTrigMode=0 - continuous sampling - SerialEncTrigEna=1 - enable trigger on - SerialEncStatusBits=2 - SerialEncNumBits=36 Two notes: 1. Renishaw documentation suggests a maximum trigger rate of 25kHz. The phase clock in your setup is 20kHz, so this should be within spec. _However_, if you look inside the included Renishaw BiSS-C mode documentation (L-9709-9005-03-F, page 3) this rate is too high to operate without line delay compensation. For prototyping could you slow the phase/servo clock and the serial encoder frequency and confirm the itntegrity of the serial communications ? 2. The CRC mask shown above doesn't appear to match the Renishaw/CRC standard. The documentation (again, L-9709-9005-03-F, page 2) shows the CRC calculation as x^6+x^1+x^0==$43, rather than $15? Here, could you check that EncTable[6].index3..6 are equal to zero? In particular, I see that if index4 !=0 then the encoder value will be integrated, per the Power PMAC Users manual, page 198: Finally, I've also attached the Excel spreadsheet that I found somewhere else on the forums, which I used to tease out the functions from the encoder control and command registers. -Jeff Dickert -Berkeley Center for Structural Biology -ALS - Beamline Controls Group L-9709-9005-03-F.PDF Technical note 40_02_CRC Calculation.pdf PPMAC_Acc84E-Biss_setup-1.xls

Does it look like the returned serial position is being integrated? If so, check EncTable[6].index4. From the Power Users Manual, page 198:

In our application, using Faraday Controls (i.e., Delta Tau UK pre-acquisition) rack-mount Power Brick hardware, we have both: - the Faraday Controls Quad PSU: http://faradaymotioncontrols.co.uk/quad-psu-power-supply/ - and a TDK-Lambda rack-mount supply with modular power supply inserts: https://uk.tdk-lambda.com/products/product-details.aspx?scid=228 Pricing was a wash between the two units, at about $2 or $3k, so relatively minor on a per-axis basis given that each could run 32 axes, if motions are sequenced so not all are running simultaneously. I missed seeing the 60VDC requirement, which nixes the Faraday Controls PSU.

My first guess would be that the absolute position read on power-cycle/reset is occurring before the encoder registers within the ACC84x have been 'filled' by a completed read from the encoder. The 'absolute home' position transferred into the motor[4]. might == 0 in that case. Moves from that point would be the relative moves you're observing. I expect that waiting a few cycles after startup to perform the absolute home read would fix this. Also, try checking the error bits up in the upper byte before allowing an absolute home; I'd think that the bits that should be '1' would also be zero in your startup condition. (BTW, we're in a similar position in handling encoder errors or loss, albeit dual Renishaw BiSS-C encoders over MACRO, so we'll need to push the error bits over MACRO I/O, after initializing and handling ring errors, etc.)

Quick note here: These instructions worked on our group's CPU board with a -106 revision, but _not_ with a -108 revision. The symptoms were that the IP address would appear to program normally (with the usual 'IP address programmed to EEPROM, do you want to set up the registry?'), but the IP address didn't change. It acted like a hardware or firmware problem. After talking with tech support, I found that the boards of the -107 and later revisions need the E1 jumper installed for the IP address upgrades. This is shown in the modern CPU/Comm board documentation where the E-point jumpers are explained, but not yet in the text that explains how the Ethernet addresses are changed.