Forums

SPI Signal or Something else

Started by Ola1a 2 weeks ago10 replieslatest reply 2 weeks ago90 views

Hello Friends, 

I am new to this Forum, so first  of all greetings to all !

I have an old Agilent system that has a detector and a DSP board. The signal is transmitted from the sensor to the board using a 40 pin flat-cable. 

I have used a logic analyzer to determine if an SPI bus is being used to transmit the signals. I can see a number of clock signals (more then 10), However not a single MOSI / MISO Signal. I have also used an Oscilloscope on this. still no luck.

Is there someone who has experience with such systems. I would be happy to pay a small fee to someone who can guide this old retired hobbiest.

Best Wishes


[ - ]
Reply by Kenneth43June 30, 2020

What type is the Agilent?

[ - ]
Reply by Ola1aJune 30, 2020

B21xB

[ - ]
Reply by jbrowerJune 30, 2020

Ola-

What is the DSP type, both mfg and part number ?  How old ?  Agilent was created in 1999 so you might have something built by HP in years prior.  Specific DSPs have different external interfaces, which can help infer the design; for example, if the DSP is an old Lucent DSP32 then we would know exactly what is the interface.

Also can you post a few pics of the boards ?

-Jeff

[ - ]
Reply by Ola1aJune 30, 2020
[ - ]
Reply by Ola1aJune 30, 2020
[ - ]
Reply by Ola1aJune 30, 2020

Post deleted by author

[ - ]
Reply by Ola1aJune 30, 2020

Dear Jeff, 


The DSP chip that I can see on the board is DSP56002FC40 (motorola inc). 


Regards

[ - ]
Reply by jbrowerJune 30, 2020

Ola-

First, what is on the sensor board ?  All 3 images you posted are the main board.

Second, what is the DSP56002 doing... It has SSI and SCI serial ports, an 8-bit HI (host interface), and an external data bus. The SSI supports "Motorola SPI compliant" devices. Near the 56002 I see an AD7884, which has a 16-bit parallel interface and samples around 160 kHz, so it may be connected to the 56002 external data bus, or it could be connected to the Xilinx FPGA which does some data reduction before the 56002 sees it.

At this point, I'm guessing the 40 pin cable is also carrying analog signals from the sensor board, which would explain why the AD7884 is so near the cable connector. To figure out what's on the 40 pin cable maybe you can do a process of elimination, first ohm out connections between it and analog devices nearest the connector (appears to be some Linear Tech and ADI amps or similar), and then between those and the AD7884. Does the cable have a lot of grounds, like every other pin is ground ?

There may well be some SPI type data transfer happening on the cable, but only if the sensor board contains devices with SPI interfaces.

-Jeff

[ - ]
Reply by Ola1aJune 30, 2020

Hello Jeff, 

Thanks for your feedback. I hope you are doing well. 

Several of the 40 pins appear to be connecting to the AD7884. In fact yes, it appears that these are all ground pins (otherwise they would not be connected).

Any yes, some of the pins that are definitely not carrying digital signals don't appear to be ground pins either. So they may be carrying some analogue signals. I have not checked all the pins to all possible analog chips nearby.

The sensor board is hidden behind an assembly that's hard to reach and I don't want to damage it. I am working on getting it open some how. 

However, I can see an EEPROM (STM 93C66W6) on the sensor board. So I went a step forward and ohmed out the 40 pins to see if I could locate the SPI pins to the EEPROM.

Yes, I have been able to determine the CS;DataIn,DataOut & Serial_Clock. 

However I am still wondering if there is any serial signal at all from the sensor board to the DSP!?

If I understand correctly, the signal is  analogue. It gets amplified using an OP27 and then fed to an ADC converter, which then throws out a 16 bit parallel digital signal...?


[ - ]
Reply by jbrowerJune 30, 2020

Ola-

There should be either one or no AD7884 pins going to the cable connector: one if they ran Vin+ from the sensor board as a single-ended input without amplification, or none if they amplified Vin+ first or ran 2 pins from the connector (Vin+ and Vin-) to a differential op-amp, which then connected to the AD7884. The 16-bit data bus + control signals would be connected to the 56002 or the large Xilinx FPGA, either directly or indirectly thru a buffer (like a '245).

To determine ground pins on the cable connector, you need to find a "known ground", for example cathode (minus) terminal on one of those D size caps near the 40-pin connector (end marked with a bar is ground). I assume you've had experience at ohming -- for non-ground signals your meter may not beep immediately, you have to hold the probe and let the caps charge up for a second or so and then you get a beep. There's a lot of caps on that board, so that's significant.

How did you determine CS, DataIn/Out and clock on the STM EEPROM ? From the data sheet ? And then you ohmed those pins to the cable connector ? 

Once you got these items ironed out, I suggest that you make a pin-out of the 40-pin header. Assuming they grounded every other pin to reduce coupling between adjacent ribbon cable signals, you should not have that many left to identify.

-Jeff