Winbond ISD1700 Voice Recorder

[mister_e]

Ship me one, and I'll do it.

[ivirscar]

Gentlemen... I'm almost there, gimme a few days

[ivirscar]

Ship me one, and I'll do it.

Me too!


Or... You can try it yourself


Most of this code was written for me by Jerson and I reconfigured to "fit my needs". The following will give you an idea on how to make the ISD1700 series jump and also telling them how high to jump AND how loud...

SCLK        var portc.0
MOSI        var portc.1
MISO        var portc.2
SS          var portc.3

' This array contains data to be sent to the ISD
' and on return it contains data reveived from ISD
ISD_Data    var byte[8]
SR0         var ISD_Data[0]
SR1         var ISD_Data[1]

isCmdErr    var SR0.0
isINT       var SR0.4
isReady     var SR1.0

Reg         var byte
Cnt         var byte
bSPI        var byte      ' the byte being transferred on SPI

    ADCON1 = %00000111    'Sets all A/D conversion to Digital input/output.
    TRISA = %111111          ' All A Ports input
    TRISC = %00000100       ' MISO is input
    TRISB = %00000000
    SS = 1
    SCLK = 1
    MOSI = 0



    ' Initialize the ISD
' ------------------------------------------------------------------------------------
    gosub PowerUp
    pause 10
    gosub WR_APC2 ' In my application, this sets the volume and config bits. 
                           'You can use it as well if you desire.
    pause 10
    gosub ClrInt

'This is the Set_Play command
SetPlay:
    isCmdErr = 1
    while isCmdErr
      ss = 0
      bSPI = $80
      gosub SPI
      ISD_Data[0]=bSPI

      bSPI = $00    
      gosub SPI
      ISD_Data[1]=bSPI

      bSPI = $10       ' Start Address low byte. (First packet of memory after effects)(Use your begin low address here)
      gosub SPI
      ISD_Data[2]=bSPI

      bSPI = $00       ' Start Address high byte (Use your begin high address here)    
      gosub SPI
      ISD_Data[3]=bSPI

      bSPI = $00       ' end address low byte (Use your end address here)
      gosub SPI
      ISD_Data[4]=bSPI

      bSPI = $00       ' end address mid byte(Use your end address here)      
      gosub SPI
      ISD_Data[5]=bSPI

      bSPI = $00       ' end address high byte / Note end address is 24 bits long. (Use your end address here)
      gosub SPI            
      ISD_Data[6]=bSPI

      bSPI = $00
      gosub SPI
      ISD_Data[7]=bSPI
      ss = 1
    wend
    return

PowerUp:
    isCmdErr = 1
    while isCmdErr
      SS=0
      bSPI=$01
      gosub SPI
      ISD_Data[0] = bSPI
      bSPI=$00
      gosub SPI
      ISD_Data[1] = bSPI
      SS=1
    wend
    return
    

'This is the command I use to set the volume as stated above. I have the factory defaults tagged in brackets for reference...

WR_APC2:
    isCmdErr = 1
    while isCmdErr
      ss = 0
      bSPI = $65
      gosub SPI
      ISD_Data[0]=bSPI

      bSPI = %01000011
      gosub SPI
      ISD_Data[1]=bSPI                      'Factory default
                                                 'high byte  'low byte
      bSPI = %00000100                   '(0100)    (01000000)
      gosub SPI
      ISD_Data[2]=bSPI
      ss=1
    wend
    return

' transact a byte with SPI interface
' You should come here after setting SS=0 for this to work
SPI:
    for Cnt=0 to 7
      MOSI=bSPI.0         ' shift MOSI from LSB

      SCLK = 0            ' clock the MOSI data
      if MISO then        ' read the MOSI data
      @ bsf status,c
      else
      @ bcf status,c
      endif
      @ rrf _bSPI         ' shift MISO into MSB
      SCLK = 1
      
    next
    return
    

' clear Interrupt and EOM bits
ClrInt:
    isCmdErr = 1
    while isCmdErr
      SS=0
      bSPI=$04
      gosub SPI
      ISD_Data[0] = bSPI
      bSPI=$00
      gosub SPI
      ISD_Data[1] = bSPI
      SS=1
    wend
    return

' read status of Chip
ReadStatus:
    SS=0
    bSPI=$05
    gosub SPI
    ISD_Data[0]=bSPI
    bSPI=$00
    gosub SPI
    ISD_Data[1] = bSPI
    bSPI=$00
    gosub SPI
    ISD_Data[2] = bSPI
    SS=1
    return

Brenon

[milestag]

I haven't tested the code yet (but thanks VERY much for posting it!!).

But I noticed one thing...

You have SR0 and SR1 as the first 2 bytes returned. The datasheet shows that the first TWO bytes returned are SR0 for every SPI command (as SR0 is 2 bytes long). So SR1 should actually be the THIRD byte returned?

SR0 var ISD_Data[0]
SR0b var ISD_Data[1] 'or whatever you want to name it
SR1 var ISD_Data[2]

Or am I missing something? I don't think this would cause a problem in your examples as you are not checking the isReady bit.

[ivirscar]

Good point... Your right, I'm not checking the ready bit, I'm simply playing a sound file at a given time (on a scanning light) and if the time approaches again before it finishes playing, the chip automatically "ignores" the next command sent UNTIL it is ready for a new command (on the next return of the light...)

But for future reference and expandability of my product, I will experiment a little on the SR0 / SR1 and correct so my code can see a ready bit in the future...


Thanks for pointing that out!


Brenon

[Jerson]

I haven't tested the code yet (but thanks VERY much for posting it!!).

But I noticed one thing...

You have SR0 and SR1 as the first 2 bytes returned. The datasheet shows that the first TWO bytes returned are SR0 for every SPI command (as SR0 is 2 bytes long). So SR1 should actually be the THIRD byte returned?

SR0 var ISD_Data[0]
SR0b var ISD_Data[1] 'or whatever you want to name it
SR1 var ISD_Data[2]

Or am I missing something?

No you are not. My mistake

Unfortunately I do not have the hardware to have noticed it and so I missed that.

[milestag]

I'm working on a modified version of the code, and hope to test it this weekend. I'll be driving the ISD1700 with a PIC16F684 and will post the final (working) code here. So far I am just checking all the commands against the datasheets.

I do have a couple observations.

The SetPlay command should only send SEVEN bytes. The END address is contained within TWO bytes (12 bits). That THIRD address byte (last byte sent) should be set to "0x00" as it is not supported in existing hardware. Same case for SetRecord.

I finally figured out your use of "While CmdError." Had me stumped for a while, then it clicked. That's pretty clever. Though it looks like in some cases it will be better to check the "RDY" status (depending on which command you are sending).

Also I did not realize that you can shift data into a byte using "status, c" carry bit in assembly. That is a very cool trick, indeed! Pretty much all of the other SPI examples I've found for PicBasic just used the Shiftout command which does not allow reading the Status bits.

For anyone interested, the original ISD1700 development kit made by Windbond (which cost me $270 from Digikey last year) does NOT support batch recording of sounds. You have to set the Start and End address manually and record one sound at a time. Very tedious and not usable other than basic function tests. However... there is a NEW ISD1700 development kit made by Nuvoton and it DOES support batch programming of wav files. So it can function as a decent production programmer. Unfortunately it's going to cost me another $270 - ouch. I called tech support at Nuvoton and they confirmed the new software is NOT compatible with the original Winbond development hardware.

You should also be aware that the Quadravox QV400D programmer never added support for the ISD1700 series. I still use it for programming ISD2560 chips, but they are starting to get expensive and much harder to find.

[mister_e]

so someone would have to make himself friend with Winbond and suggest them an universal Gang device programmer

Not sure this worth it, look at current storage device pricing and add a higher level Microcontroller to it... NO way you're going to be over the Price of any single Winbond IC... not to mention you can update any part of your firmware and sound using a simple bootloading process.

Roman Black sounds almost the same as most Winbond solution and it's free, MP3 codec ICs are cheap, DsPIC are also cheaper... food for thought.

[Normnet]

Not sure this worth it, look at current storage device pricing and add a higher level Microcontroller to it... NO way you're going to be over the Price of any single Winbond IC... not to mention you can update any part of your firmware and sound using a simple bootloading process.

Requires porting of basic code but take a look at PIC iPod wav player.

Mono CD quality (16bit 44k) for the price of an SD card and holder,
a PIC18F4520, a few resistors for an R2 ladder, an audio amp and a speaker.

Or stereo 8bit 44k if preferred.
Up to 2 Gig's or 512 wav files.


Norm

[milestag]

so someone would have to make himself friend with Winbond and suggest them an universal Gang device programmer

Yes, Winbond (Nuvoton?) does/did offer a production gang programmer. 12 chips at once I believe. But cost is around $2000

[ivirscar]

Jerson done a FANTASTIC job while not having any hardware to test with!! These ISD chips are difficult to work with at a glance anyways...

I done more changing to the code I'm using by adding the 3rd byte to the status register and "singling out" the 2nd bit to check the status of the PLAY mode. With this bit, I can control the amplifier, which sound to play and when and so on...

More than adequate for the application I have

[txniteowl]

I'm working on a modified version of the code, and hope to test it this weekend. I'll be driving the ISD1700 with a PIC16F684 and will post the final (working) code here. So far I am just checking all the commands against the datasheets.

I do have a couple observations.

The SetPlay command should only send SEVEN bytes. The END address is contained within TWO bytes (12 bits). That THIRD address byte (last byte sent) should be set to "0x00" as it is not supported in existing hardware. Same case for SetRecord.

I finally figured out your use of "While CmdError." Had me stumped for a while, then it clicked. That's pretty clever. Though it looks like in some cases it will be better to check the "RDY" status (depending on which command you are sending).

Also I did not realize that you can shift data into a byte using "status, c" carry bit in assembly. That is a very cool trick, indeed! Pretty much all of the other SPI examples I've found for PicBasic just used the Shiftout command which does not allow reading the Status bits.

For anyone interested, the original ISD1700 development kit made by Windbond (which cost me $270 from Digikey last year) does NOT support batch recording of sounds. You have to set the Start and End address manually and record one sound at a time. Very tedious and not usable other than basic function tests. However... there is a NEW ISD1700 development kit made by Nuvoton and it DOES support batch programming of wav files. So it can function as a decent production programmer. Unfortunately it's going to cost me another $270 - ouch. I called tech support at Nuvoton and they confirmed the new software is NOT compatible with the original Winbond development hardware.

You should also be aware that the Quadravox QV400D programmer never added support for the ISD1700 series. I still use it for programming ISD2560 chips, but they are starting to get expensive and much harder to find.

--------
I found a Development Board for $70 from Elusions Co.

Dennis

[RFEFX]

Thanks everyone (including Jerson) for the code for these ISD devices. However we only have half of the issue resolved.

According to the code only "set_play" was included. How you gonna play something if it is not recorded?

What i am getting at is... in my original post.....

Show me a simple code to Record 5 messages using set_rec and play any of them back using Set_play... i.e. for the ISD17120

Thanks again.

[Ioannis]

Trying to understand how these devices work, I could not found a way to play e.g. the third message out of 10. How is the messages indexed? I mean if you record 10 messages, how can you know the position of each and how to play them?

Are these devices confussing or am I too dumb to get it? Doh...

Ioannis

[midali]

I found the problem ..

TRISC = %00000000

Now all is ok !