Post by thread:[PIC] SPI ISDs and PIC

> Does anyone have code examples for the SPI ISD chips, in my > case the ISD5008. I'm upgrading a product from the smaller ISDs > which use the A0-7 interface. Working through the manual but > would like to see how it's been done before and also any gotchas > > For example, 01110000 is given as the instruction code for both > STOP and RINT. Might be true, still browsing. Can't see any > appnotes or errata at Winbond > > TIA > Here's some bit banged SPI ISD code used in the http://www.brailletutor.com ;------------------ISD Speech Chip Routines--------------------------------------------- ISDPlayTxMidLo ; Load TxMid/TxLo into ISD, then play it call ISDsetPlay goto ISDplay ISDpowerUp ; Send power up command to ISD chip. if isd==4003 movlw b'00100000' ; Power up command 0f 00100 with lower bits don't care movwf TxMid goto SPI endif if isd==4004 movlw 0x20 movwf TxHi goto SPIsendHi ; Send just the high byte endif ISDsetPlay ; Set the play address. if isd==4003 ;Preload the 11 bits of address in TxMid/TxLo. movfw TxMid ; Get high address andlw 0x07 ; Clear bits higer than valid range addlw b'11100000' ; Add in set play command movwf TxMid goto SPI endif if isd==4004 ; Just set TxHi, leaving TxMid and TxLo alone movlw 0xe0 movwf TxHi goto SPI ; Send 3 bytes endif ISDplay ; Start playing at address set by ISDsetPlay. Stop at EOM or overflow ; call WaitEOM ISDplayNoWait if isd==4003 movlw b'11110000' movwf TxMid goto SPI endif if isd==4004 movlw 0xf0 movwf TxHi goto SPIsendHi ; Send just the high byte endif ISDstop ; Stop the current action. if isd==4004 movlw 0x03 movwf TxHi goto SPisendHi ; Send and return endif return ; Just return if ISD4003 ISDpowerDown ; Shut the chip down. if isd==4003 movlw b'00010000' movwf TxMid goto SPI endif if isd==4004 movlw 0x01 movwf TxHi goto SPIsendHi endif WaitEOM ; Wait for EOM from ISD chip. Includes a timeout in case it's missed cblock eomLo eomMid eomHi ; Timers for timeout endc clrf eomLo clrf eomMid clrf eomHi ; Clear the timer WaitEOMa btfss portb,0 ; See if ISD has sent an interrupt (indicating EOM) return ; Got it, exit decfsz eomLo,1 goto WaitEOMa ; Go around again clrwdt decfsz eomMid,1 goto WaitEOMa ; To around again unless we timed out decfsz eomHi,1 goto WaitEOMa ; To around again unless we timed out return ; Give up waiting for EOM SPI ; Performs an SPI data exchange with the ISD4003. ISD4003 uses LO and MID registers (16 bits). ISD4004 uses 24 bits cblock TxLo ; Low byte to be transmitted to ISD4003 TxMid TxHi TxRot ; Rotate bits out of this register to avoid destroying Tx registers RxLo ; Low byte received from ISD4003 RxMid RxHi RxRot ; Rotate bits into this register as received SPIbitCount endc SPInoWait bcf lata,3 ; -SS low, starting cycle movfw TxLo ; Get low byte to transmit movwf TxRot ; and get ready to shift it out call DoSPIbyte ; Go send it movfw RxRot movwf RxLo ; Save the received byte movfw TxMid movwf TxRot ; Get mid byte to rotate out call DoSPIbyte movfw RxRot movwf RxMid ; Save received byte ; more stuff here for 4004 if isd==4004 SPIsendHi ; Late entry point to send just one byte (TxHi) bcf lata,3 ; -SS low (if not already low) movfw TxHi movwf TxRot ; Get hi byte to rotate out call DoSPIbyte movfw RxRot movwf RxHi ; Save received byte endif bsf lata,3 ; -SS high return DoSPIbyte ; Send byte in TxRot and receive byte into RxRot movlw 0x08 movwf SPIbitCount ; Initialize the bit counter for low byte SPIa clrc ; Assume bit is low btfsc porta,1 setc ; Set carry if incoming data bit was high rrf RxRot,1 ; Rotate into receive register rrf TxRot,1 ; Rotate first bit to transmit into carry skpc bcf lata,0 ; Bit was clear, so clear output bit skpnc bsf lata,0 ; Bit was set, so set output bit nop ; Let output port settle bsf lata,2 ; Clock line high nop bcf lata,2 ; Clock line low nop decfsz SPIbitCount,1 ; Loop 'til 8 bits sent and received. goto SPIa return Harold -- FCC Rules Updated Daily at http://www.hallikainen.com - Advertising opportunities available!

Harold, what voltage do you run the ISD400x at ? Although the specs say 2.7V to 3.3V, I notice the absolute max is 7V, which seems like a high enough ceiling to use 4-5V, more in line with the PIC and the rest of the 5V system. Perhaps maybe power it via a diode for a bit of a drop rather than have another regulator. Which would be the desirable thing to do, it just seems the margin between recommended and absolute seems rather wide One reason I ask is that a series of receiver/transmitter pairs I use are also specced at 3V/5V in the datasheet, but in fact will work at up to 12V. For some reason, probably to emphasise that they can work satisfactorily with a couple of batteries, only 3V/5V is ever mentioned in the original, sketchy, datasheet, and that 3V/5V has been quoted ever after in other datasheets

our design ran the PIC at 5V, and the Winbond at 3.3 and the audio amp at 12V. Jinx <spam_OUTjoecolquittTakeThisOuTclear.net.nz> wrote: > I ran it at 3V. I've posted the schematic at > http://sujan.hallikainen.org/temp/schematic.pdf Thanks. Although the 5008 has more internal modules, it's similar in basic function to the 400x. I've had one sitting around here for a couple of years. It may have come from Russell. Turned out to be handy having one to hand anyway > I haven't looked at the datasheet in a while. I don't recall the > 7V absolute max. That's pretty high! It is when you compare it with the 3.6V max for an LF PIC. 7V seems to me to be more in line with a 5V Vcc If I had two of the 5008 I'd be trying it at 5V. For now, 3V. I'll drop a line to Winbond, see if they have any comment

> I haven't looked at the [4003] datasheet in a while Been struggling to get this 5008 working properly. Don't know if anyone's working with these at the moment, but just to put this out there for future reference - Comparing the 400x and 5008 datasheet, I noticed that the 400x procedure has an additional POWERUP command before a REC command. The 5008 manual says one POWERUP will enable the chip. Doesn't seem to. I added another POWERUP and it came right. All in all, the ISD datasheets could be (should be) better For the power supply, 3 x IN914 diodes from PIC's 5V with a 4k7 load keeps the ISD Vcc around 3V. Power-down not required on this project. ISD could be powered or enabled by the PIC though Celebrated, in my own humble way, with a recording and playback of Ian Dury's "There Ain't Half Been Some Clever Bastards" ;-)) clr_ss ;/SS = 0 + 100us delay mov b'01100000',data_out ;POWERUP call data8 ;send data set_ss ;/SS = 1 + 100us delay call ms50 ;50ms delay (Tpu > 25ms) ;set configs for record cfg_rec mov b'00001011',cfg0_lo ;0x440b mov b'01000100',cfg0_hi mov b'11000001',cfg1_lo ;0x00c1 mov b'00000000',cfg1_hi call set_cfgs record prtstr2 recording ;"Recording" clr_ss ;/SS = 0 + 100us delay mov b'01100000',data_out ;POWERUP call data8 ;send data set_ss ;/SS = 1 + 100us delay call ms100 ;100ms delay (2 x Tpu > 25ms) clr_ss clrf data16_1 ;address = 0 clrf data16_2 call data16 ;send data usec10 movlw b'10100000' ;record from set address movwf data_out call data8 set_ss

> Congratulations on your success! Thanks, it's a great relief to see it doing what the manual says it should. I thought I might have to go out and buy some frowns and sighs wholesale !! > I never tried to directly record on the ISD chips. I created a > BUNCH of audio files, then used a Quadravox programmer > to program the chips. The Quadravox software gives you a list > of the addresses each recording starts at. So, I just had a table > of address I sent the chip Part of this project is a message recorder for customers, kind of voice mail, so I really had to get into the nitty-gritty of the chip. The addition of INT and RAC indicators is very helpful, although CUEing to EOM markers will probably do for mail playback at this stage. I'd have to do a comparison but to my ears the quality sounds better than the 1400, although the datasheet THD is the same I'll put it up as a page when it's all sorted out

> Congratulations on your success! Still looking into this. Found a PIClist post from several years ago in which the solution was also double-instructing, this time the user was having trouble with playback. David Duffy replied to this post, appearing to say he has no trouble with the 5008, with code examples Might get Winbond to comment on what could be a recognised glitch. The interface is pretty straight-forward, might be a timing thing in the chip itself http://techref.massmind.org/techref/piclist/2001/05/02/193409a.txt > I resolved yesterday the code, all what I write > some times ago,it was right but this chip need > 2 successive access in control registers when > the settings will be actives. I send 2 successive > play commands and the chip play the message until > find the EOM. > thanks again. > Miki

part 1 2420 bytes content-type:text/plain; (decoded 7bit) > I never tried to directly record on the ISD chips. I created > a BUNCH of audio files, then used a Quadravox programmer I may get a programmer, but for now I've kind of automated the process. The 25 messages are assembled on the right channel of Goldwave. Separating them on the left channel are tone bursts. These are fed into a 567 tone decoder and its output tells the PIC what data and when to send it to the ISD (The audio in the picture is actually Dick Clark saying "Two". I found a New Year countdown mp3 of him and Ed McMahon which was very useful for testing whether all the messages had stuck and if they could be recalled correctly) If the PIC powers up with button1 down, it enters Record mode. kb_567 looks for a manual button in this mode ;Pushbuttons in Record mode ;Diode-steered into high nybble of PortD ;Play at 00 button1 ;Play at current address button2 ;Record at 00 button3 ;Record at current address button4 ;Stop button5 ;Cue to next message button6 ;Exit button7 ;spare button8 kb_567 movfw portd swapf wreg andlw b'00001111' skpz ;no button pressed bra rec_234 ;button down, decode and release lp567 btfss ne567 ;rising edge bra kb_567 ;no tone prtstr2 stopped ;print "Stopped" clr_ss movlw b'01110000' ;stop command movwf data_out call data8 set_ss call ms100 btfsc ne567 ;falling edge bra $-2 btfsc first ;first message, record at 00 bra recatxx ;not first bsf first recat00 prtstr2 record00 ;print "Record at 00" clr_ss clrf data16_1 ;address = 0 clrf data16_2 call data16 usec10 movlw b'10100000' ;record from set address movwf data_out call data8 set_ss bra lp567 recatxx prtstr2 recordxx ;print "Record at xx" clr_ss movlw b'10110000' ;record from current address movwf data_out call data8 set_ss bra lp567 part 2 3021 bytes content-type:image/gif; (decode) part 3 35 bytes content-type:text/plain; charset="us-ascii" (decoded 7bit)

Been having quite the discussion with ISD engineers over the past few days and have resolved some issues Firstly, I have to congratulate them on being very receptive to my queries. Their response has been excellent (a pity I needed to raise the points in the first place but .....). A certain company dear to our hearts would not say > Hi Jinx, Thanks a lot for your precious comments Rather "Oh sure. Yeah, we'll get right on that". Not > I found another revision of datasheet. Do you think this one > is better than the one you have? If yes, I will proceed the > modification on this revision The isd.com pdf for the ISD5008 is dated August 2000. The person I've been corresponding with has discovered another datasheet (as above) dated March 2004. As he says "There is no link for this datasheet yet. I just found it in one of our design engineers' folder. I will discuss with our design engineers, modify the datasheet, send it to you, and put it to our website" [Jinx] The 5008 datasheet says that an interrupt will be seen on INT when a STOP command is sent. I don't see these interrupts. The phrases are being recorded and INT goes low at an EOM, but not at STOP [Eden] According to our design engineer, STOP does not cause the INT to go low. May I know where in datasheet you see the description so we can correct it ? The new pdf he found doesn't mention STOP sending INT low, so that was fixed in that (as yet unreleased) revision [Jinx] Does the 5008 have a SETMC command like the 4003 does ? I have over 20 phrases in a menu system and will need to go backwards as well as forwards depending on responses. It seems to me the only way is a PLAY @ 0 command followed immediately by a STOP and then Cue forwards. I wonder why the 5008 does not have a SETMC in the datasheet [Eden] Yes the 5008 does not have a SETMC command. I have to check with another design engineer about the reason we took it off. I will respond to you later As it turns out, the 5008 DOES have a SETMC command. I tried the 4003 SETMC, adjusted to 24-bit for the 5008, and it works. It's just been left out of both 5008 manuals for apparently no good reason rst_ptr call power_up call ms60 ;Tpud * 2, 60ms delay (8kHz sampling) clr_ss clrf data16_1 ;address = 0 clrf data16_2 call data16 ;send address usec10 movlw b'11101000' ;SETMC movwf data_out call data8 ;send command set_ss The 4003 description of SETMC says "Initiates Message Cueing from address <A0-A10>" The SETMC command I've tried for the 5008 seems to do just that, if I'm interpreting it correctly. It resets the address to 0 and then performs a Cue to the start of message 2. (I've not used the 4003 so I can't say for sure what SETMC does on it). Play @ 0 is available for message 1 of course A couple of things I've asked them to make clearer and more consistent with the 4003 manual [Jinx] The 4003 datasheet says that a second POWERUP command is needed, but the 5008 datasheet doesn't mention this. [snip] Can you clarify that please ? [Eden] According to our design engineer, a second POWERUP is needed [Jinx] The 5008 manual does not mention pull-up resistors for RAC and INT and they aren't in Fig21. Pull-up resistors for RAC and INT are mentioned in the 4003 manual descriptions, and also shown in Fig9. Those not familiar with pull-ups might not be aware of their purpose [Eden] Thank you for that suggestion, it would help