[WP-34S] Cable and flashing options for Mac users

[pascal_meheut]

(01-18-2015 09:56 PM)Marcus von Cube Wrote:  I've had the best results with chips/cables from FTDI and the Windows based MySamba software.

Same here even though I wrote the wp34sflash program and it should work the same being rewritten from the MySamba source code and sending the exact same bytes.
Flashing is incredibly sensitive to the driver and other parameters and I have never been able to make our "portable" flash program work as well as MySamba (and never and the time nor the inclination to do so).

[John Galt]

Thanks for the replies, much appreciated. I am traveling for a few days and will report what I find later.

(01-18-2015 09:56 PM)Marcus von Cube Wrote:  It therefore sends some trigger bytes at what it assumes to be 115kbit/s. It then gets back an answer (or not) and retries with a slightly modified oscillator setting. Maybe the Prolific chips just discard the sync data if the baud rate is off too far. So the Atmel boot loader never knows how to adjust the frequency because it never sees any response from the PC or Mac.

Interesting. What a mess! I was curious what baud rate was used, especially in a device lacking a reference oscillator. In one of my attempts I tried sending the binary file using Markus Schmid's ZOC at 9600 baud. It repeatedly stalls at about 31% after which his program froze, leaving the open port in an unknown state. It's an otherwise excellent program, but things like this leave me skeptical.

I'll try again at 115 kbps but if MySamba (and wp34sflash) are required to provide intelligent responses to queries from the WP34s then I obviously can't simply use a terminal emulator to flash the unit.

Which leads me to:

(01-18-2015 10:42 PM)jebem Wrote:  you could test your PC to USB-serial-cable link, by connecting the TxD and RxD wires with one 1K resistor to from a loopback.
On your tty test program, you should receive every single character that you type on the keyboard.

Yes, I did that, and the initial results using ZOC led me to believe the loopback was not working. I was going to measure the serial signal with an oscilloscope but I had little time for further testing. I'll do that, and try other TTY - emulator software when I return. It would be nice to confirm actual data transmission, or at least the attempt to send data.

(01-19-2015 05:50 AM)pascal_meheut Wrote:  ... I wrote the wp34sflash program and it should work the same being rewritten from the MySamba source code and sending the exact same bytes.
Flashing is incredibly sensitive to the driver and other parameters and I have never been able to make our "portable" flash program work as well as MySamba (and never and the time nor the inclination to do so).

Thank you for writing that program! Do you know of any USB to serial devices that would work for my application? The cable I bought might be unsuitable, perhaps even inoperative, but it only cost $2 to find out.

[Marcus von Cube]

(01-19-2015 06:50 PM)John Galt Wrote:  In one of my attempts I tried sending the binary file using Markus Schmid's ZOC at 9600 baud. It repeatedly stalls at about 31% after which his program froze, leaving the open port in an unknown state. It's an otherwise excellent program, but things like this leave me skeptical.

I'll try again at 115 kbps but if MySamba (and wp34sflash) are required to provide intelligent responses to queries from the WP34s then I obviously can't simply use a terminal emulator to flash the unit.

No chance! The SAM-BA boot loader knows some commands to modify RAM and start code. That's what both MySamba and the Atmel SAM-BA software do: They download code to RAM and then let it do all the "dirty work". The main difference between MySamba (and its siblings) and SAM-BA from Atmel is the downloaded code. The smart HP people (I assume mainly Cyrille) just wrote a flashing algorithm that is a factor 10 faster than what Atmel provides.

[Dave Frederickson]

(01-19-2015 06:50 PM)John Galt Wrote:  Do you know of any USB to serial devices that would work for my application? The cable I bought might be unsuitable, perhaps even inoperative, but it only cost $2 to find out.

I opted for the easy way out and bought a cable from Harald.

[John Galt]

(01-19-2015 07:25 PM)Dave Frederickson Wrote:  I opted for the easy way out and bought a cable from Harald.

You mean the USB board that installs inside the calculator? I know that's an option but I'm trying to avoid additional modifications and use only the externally accessible programming port. Is it possible to use his board in that manner?

Or are you describing something else? If so I'm not familiar with it.

[Dave Frederickson]

(01-19-2015 07:40 PM)John Galt Wrote:  

(01-19-2015 07:25 PM)Dave Frederickson Wrote:  I opted for the easy way out and bought a cable from Harald.

You mean the USB board that installs inside the calculator?

No, the USB-Serial cable assembly also sold by Harald described in the WP 34S manual. It includes the two buttons.

Dave

[pascal_meheut]

(01-19-2015 06:50 PM)John Galt Wrote:  Thank you for writing that program! Do you know of any USB to serial devices that would work for my application? The cable I bought might be unsuitable, perhaps even inoperative, but it only cost $2 to find out.

When I was working on it, the only one working reliably on OSX (sometimes after several retries) was the KeySpan USA19HS.

[John Galt]

(01-19-2015 08:24 PM)pascal_meheut Wrote:  When I was working on it, the only one working reliably on OSX (sometimes after several retries) was the KeySpan USA19HS.

Great! I have one of them. I'll try it.

It's the same one I use to communicate with the 48GX. As a matter of fact, I'll try sticking resistors in the end of the cable that connects to the 48GX. That's easy enough!

[John Galt]

(01-19-2015 08:12 PM)Dave Frederickson Wrote:  No, the USB-Serial cable assembly also sold by Harald described in the WP 34S manual. It includes the two buttons.

Two blue buttons? I was under the impression the one in the picture was the one no longer available and not even worth looking for (see my initial post in this thread). If Harald sells them I'll investigate that. It would certainly be convenient.

My cursory attempts to find Harald, his USB board, and – now that I know about it – this mysterious cable have not been successful. That's all part of the fun, but if I won't mind if you give me a hint.

:-)

[Dave Frederickson]

Hi John,

No, the blue-button cable that was originally available from Gene was RS-232 - you still needed a USB-232 adapter.

Harald's boards are well documented in the WP 34S manual. He can be emailed at hdotpottatgmxdotnet (make the appropriate substitutions) or PM'd here.

Dave

[Dave Frederickson]

I apologize. Only one of Harald's boards is documented in the WP 34S manual. The other's are described here.

[John Galt]

(01-19-2015 06:50 PM)John Galt Wrote:  

(01-18-2015 10:42 PM)jebem Wrote:  you could test your PC to USB-serial-cable link, by connecting the TxD and RxD wires with one 1K resistor to from a loopback.

Yes, I did that, and the initial results using ZOC led me to believe the loopback was not working.

Follow up: my initial results were incorrect - the loopback test definitely works, so for now I have to conclude wp34sflash is not compatible with the Prolific device.

(01-18-2015 09:56 PM)Marcus von Cube Wrote:  Maybe the Prolific chips just discard the sync data if the baud rate is off too far. So the Atmel boot loader never knows how to adjust the frequency because it never sees any response from the PC or Mac.

Just guessing, I'm afraid.

I believe your "guess" is 100% correct.

[John Galt]

(01-19-2015 08:24 PM)pascal_meheut Wrote:  When I was working on it, the only one working reliably on OSX (sometimes after several retries) was the KeySpan USA19HS.

This is a long shot but - since I given up for now trying to make the Prolific device work - have you ever had occasion to test your program with FTDI's FT230X device? It's the one used in Harald's board.

[Massimo Gnerucci]

I always used devices with Prolific chips with perfect results.
However you should be sure to have true Prolific chipsets, and not Chinese lookalikes pretending to be the true thing.

[John Galt]

(01-20-2015 07:05 PM)Massimo Gnerucci Wrote:  I always used devices with Prolific chips with perfect results.

To be clear, are you also using pascal_meheut's wp34sflash program on a Mac?

I have no doubt the Prolific chip will work, since plenty of people have reported success, but without a Mac flashing utility functionally equal to "MySamba" it seems I can't use it.

FWIW I disassembled the USB housing and inspected the chip. It bears what appear to be authentic Prolific markings, Prolific's supplied OS X driver works perfectly, the signal levels measured by an oscilloscope are correct, the waveforms are perfectly square, so despite the fact it only cost $1.79 including shipping from China, I have no reason to believe it's a fake.

[pascal_meheut]

(01-20-2015 06:34 PM)John Galt Wrote:  This is a long shot but - since I given up for now trying to make the Prolific device work - have you ever had occasion to test your program with FTDI's FT230X device? It's the one used in Harald's board.

Yes. It worked fine on Windows & Linux but not on the Mac. It depends a lot on the driver though.
I really recommend using Windows from Parallels, VMWare, VirtualBox or Bootcamp.

[dewster]

I recommend you move this to a PC. The WP-34S programming process is opaque enough - add yet one more unknown to the chain and you will likely go insane.

Serial ports are a shaky way to do this kind of thing, no flow control, no error correction, just a data blast with a checksum or CRC or something at the very end that - if you are lucky - lets you know if you blew it. When it works it works, but god help you if something goes wrong.

[Dave Frederickson]

(01-20-2015 07:58 PM)John Galt Wrote:  It bears what appear to be authentic Prolific markings, Prolific's supplied OS X driver works perfectly, the signal levels measured by an oscilloscope are correct, the waveforms are perfectly square, so despite the fact it only cost $1.79 including shipping from China, I have no reason to believe it's a fake.

It's well known that counterfeit chips exist as dewster discovered last year. See Post #55 in this thread. In Windows, Device Manager can be used to inspect the VID/PID hardware ID's to identify the chip manufacturer.

[John Galt]

(01-20-2015 09:12 PM)Dave Frederickson Wrote:  It's well known that counterfeit chips exist as dewster discovered last year.

Thanks... I am familiar with Prolific's warning letter on the subject and have no reason to believe the device I'm using is counterfeit.

OS X identifies it as:

Code:

USB-Serial Controller:

  Product ID:    0x2303
  Vendor ID:    0x067b  (Prolific Technology, Inc.)
  Version:    3.00
  Speed:    Up to 12 Mb/sec
  Manufacturer:    Prolific Technology Inc.
  Location ID:    0xfd140000 / 6
  Current Available (mA):    500
  Current Required (mA):    100

This site verified the VID/PID extracted from the device and reported above are correct. If it's a fake, it's such a good one I think the only means of identifying it as such would require forensic methods beyond my reach.

[John Galt]

(01-20-2015 08:48 PM)dewster Wrote:  - add yet one more unknown to the chain and you will likely go insane.

Well, I'm already there so no loss!

HP totally abandoned the Mac platform years ago, one of many poor decisions IMO. My goal is to develop a Mac-only solution, without reliance on Windows, so that the user community can benefit from this marvelous project. Lacking absolute assurance from someone who has verified the Prolific device works with wp34sflash and OS X, I believe pursuing a Mac flashing option with this particular device will be futile.