Cheap Shenzhen Inspection Microscopes

photo 5

Well I finally got round to playing with the inspection microscopes I picked up in Shenzhen. If I remember correctly the whole system came to about 80USD. The purchasing process is almost as exciting as the price, you need to buy the optics from one place, the CCD PCBs from another, and the CCD chassis from yet another place. Luckily they’re all in the same building which specializes in security equipment.

Here’s a selection of parts:

photo 1

photo 2

The CCD cameras and optics are all C-mount sometimes with C to CS (?) adapters. This is handy as most modern microscopes use C-mount and for older microscopes you can get eyepiece to C-mount adapters.

photo 3

There are various CCD and CMOS imaging PCBs available. With a variety of network, composite, USB, and HDMI connections. Interesting composite seems to be favored for its low latency. I guess this is particularly important for inspection and rework. However it’s a little sad that the digital stuff isn’t faster these days…

I’ve been using the CCD, mostly because I’m kind of interested in CCD interfacing. It module uses a Sony CCD and CCD driver IC. It takes a 12v input. I hacked the connector off and used a lab PSU the supply it. Connecting the black to ground and other two wires to 12v. The module draws about 70mA. I used an old composite input monitor I had. I think I’ll buy some small monitors and a PC composite input acquisition card of some kind.

I purchased 2 sets of optics. One was fixed magnification, the other variable. First the fixed, the pads shown are for an SOIC8:

photo 1(1)

photo 5

Then the variable, at max magnification. This is one of the pads from the device above:

photo 3(1)

photo 2(2)

As you can see the first image I took was pretty grey and washed out. This was due to poor lighting. Without illumination the images from the variable magnification lens were much poorer. However I’m very impressed by the working distance should be great for rework.

One issue with the variable magnification lens is that you have to keep refocusing as you change magnification. This isn’t an issue with the Wild Heerbrugg M8 which I’ve been using for rework lately, and it’s very handy to be able to zoom in and out. That however has a secondhand value for a 1000USD+.

AD5791 Board Patches Revisited

photo(3)

Tried to tidy up the patches a bit and fix the gain issue previously noted appears to have been a config error (wasn’t configured correctly for the external unity gain opamp). Test code follows:

#include <SPI.h>

const int reset = A5;
const int clr   = A4;
const int ldac  = A3;
const int sync  = A2;

#define AD5791_NOP 0 // No operation (NOP).
#define AD5791_REG_DAC 1 // DAC register.
#define AD5791_REG_CTRL 2 // Control register.
#define AD5791_REG_CLR_CODE 3 // Clearcode register.
#define AD5791_CMD_WR_SOFT_CTRL 4 // Software control register(Write only).

typedef enum {
ID_AD5760,
ID_AD5780,
ID_AD5781,
ID_AD5790,
ID_AD5791,
} AD5791_type;

struct ad5791_chip_info {
unsigned int resolution;
};
static const struct ad5791_chip_info ad5791_chip_info[] = {
[ID_AD5760] = {
.resolution = 16,
},
[ID_AD5780] = {
.resolution = 18,
},
[ID_AD5781] = {
.resolution = 18,
},
[ID_AD5790] = {
.resolution = 20,
},
[ID_AD5791] = {
.resolution = 20,
}
};

AD5791_type act_device;

/* Maximum resolution */
#define MAX_RESOLUTION 20
/* Register Map */
#define AD5791_NOP 0 // No operation (NOP).
#define AD5791_REG_DAC 1 // DAC register.
#define AD5791_REG_CTRL 2 // Control register.
#define AD5791_REG_CLR_CODE 3 // Clearcode register.
#define AD5791_CMD_WR_SOFT_CTRL 4 // Software control register(Write only).
/* Input Shift Register bit definition. */
#define AD5791_READ (1ul << 23)
#define AD5791_WRITE (0ul << 23)
#define AD5791_ADDR_REG(x) (((unsigned long)(x) & 0x7) << 20)
/* Control Register bit Definition */
#define AD5791_CTRL_LINCOMP(x) (((x) & 0xF) << 6) // Linearity error compensation.
#define AD5791_CTRL_SDODIS (1 << 5) // SDO pin enable/disable control.
#define AD5791_CTRL_BIN2SC (1 << 4) // DAC register coding selection.
#define AD5791_CTRL_DACTRI (1 << 3) // DAC tristate control.
#define AD5791_CTRL_OPGND (1 << 2) // Output ground clamp control.
#define AD5791_CTRL_RBUF (1 << 1) // Output amplifier configuration control.
/* Software Control Register bit definition */
#define AD5791_SOFT_CTRL_RESET (1 << 2) // RESET function.
#define AD5791_SOFT_CTRL_CLR (1 << 1) // CLR function.
#define AD5791_SOFT_CTRL_LDAC (1 << 0) // LDAC function.
/* DAC OUTPUT STATES */
#define AD5791_OUT_NORMAL 0x0
#define AD5791_OUT_CLAMPED_6K 0x1
#define AD5791_OUT_TRISTATE 0x2

void setup() {
  Serial.begin(115200);

  SPI.begin();
 
  pinMode(11,OUTPUT);
  digitalWrite(11,LOW);

  pinMode(reset, OUTPUT);
  pinMode(clr  , OUTPUT);
  pinMode(ldac , OUTPUT);
  pinMode(sync , OUTPUT);  

  // setup
  digitalWrite(ldac,HIGH);
  digitalWrite(reset,HIGH);
  digitalWrite(clr,HIGH);
  digitalWrite(sync,HIGH);
  SPI.beginTransaction(SPISettings(20000000, MSBFIRST, SPI_MODE1));
    
  long status = AD5791_GetRegisterValue(AD5791_REG_CTRL);
  
  status = AD5791_GetRegisterValue(AD5791_REG_CTRL);
  
  Serial.print(status,BIN);
  Serial.println();

  digitalWrite(ldac,LOW);
  status = AD5791_SetRegisterValue(AD5791_REG_DAC, 1);
  digitalWrite(ldac,HIGH);


  unsigned long oldCtrl = status;
  oldCtrl = oldCtrl & ~(AD5791_CTRL_LINCOMP(-1) | AD5791_CTRL_SDODIS | AD5791_CTRL_BIN2SC | AD5791_CTRL_OPGND);

  oldCtrl = oldCtrl | AD5791_CTRL_RBUF;
  status = AD5791_SetRegisterValue(AD5791_REG_CTRL, oldCtrl);
  
  status = AD5791_GetRegisterValue(AD5791_REG_CTRL);
  


  Serial.print(status,BIN);
  Serial.println();
  Serial.println("DAC");

  digitalWrite(ldac,LOW);
  status = AD5791_SetRegisterValue(AD5791_REG_DAC, 1);
  digitalWrite(ldac,HIGH);

}



long AD5791_SetRegisterValue(unsigned char registerAddress, unsigned long registerValue) {
  unsigned char writeCommand[3] = {0, 0, 0};
  unsigned long spiWord = 0;
  char status = 0;
  spiWord = AD5791_WRITE | AD5791_ADDR_REG(registerAddress) | (registerValue & 0xFFFFF);
  writeCommand[0] = (spiWord >> 16) & 0x0000FF;
  writeCommand[1] = (spiWord >> 8 ) & 0x0000FF;
  writeCommand[2] = (spiWord >> 0 ) & 0x0000FF;
  
  digitalWrite(sync,LOW);
  status = SPI.transfer(writeCommand[0]);
  status = SPI.transfer(writeCommand[1]);
  status = SPI.transfer(writeCommand[2]);
  digitalWrite(sync,HIGH);

  return 0;
}

long AD5791_GetRegisterValue(unsigned char registerAddress) {
  unsigned char registerWord[3] = {0, 0, 0};
  unsigned long dataRead = 0x0;
  char status = 0;
  registerWord[0] = (AD5791_READ | AD5791_ADDR_REG(registerAddress)) >> 16;

  digitalWrite(sync,LOW);

  status = SPI.transfer(registerWord[0]);
  status = SPI.transfer(registerWord[1]);
  status = SPI.transfer(registerWord[2]);
  digitalWrite(sync,HIGH);


  registerWord[0] = 0x00;
  registerWord[1] = 0x00;
  registerWord[2] = 0x00;
    digitalWrite(sync,LOW);
  registerWord[0] = SPI.transfer(0x00);
  registerWord[1] = SPI.transfer(0x00);
  registerWord[2] = SPI.transfer(0x00);
    digitalWrite(sync,HIGH);
  dataRead = ((long)registerWord[0] << 16) |
             ((long)registerWord[1] << 8) |
             ((long)registerWord[2] << 0);
  return dataRead;
}

long current_value=0;
long d=0;
void loop() {
   
    digitalWrite(ldac,LOW);
    int status = AD5791_SetRegisterValue(AD5791_REG_DAC, current_value);
    digitalWrite(ldac,HIGH);
   // Serial.print(current_value);
   // Serial.print("\r\n");
 
  current_value+=10000; 
if(current_value>524000) current_value=-524000;
}

AD5791 Arduino Shield Rev 2

ad5791_rev2

Updated the AD5791 board with the revisions from yesterday and sent it to fab.

Kicad files: ad5791.tar

AD5791 Shield First Rev

 

photo(2)I received the AD5791 Arduino shield PCBs from the fab today. And built them up. There are a number of revisions that I’ll need to make:

  • The holes on the SMA connector are too small.
  • AGND and DGND are not tied near the AD5791
  • The Pads on U1,U2 and U3 are too far apart for my liking.
  • U3 has a via on a pad which I don’t like
  • The Tantalum caps should have polarity markings.
  • U5 silk marking is unclear (actually a lot of the silk is unclear)
  • C23 footprint is wrong cap type
  • C5 has the wrong polarity in the schematic
  • pin 7 on U1 is NC. It should be connected to the feedback resistor. I blame Kicad.
  • pin 5&6 and 2&3 on U3 are the wrong way round (can’t blame Kicad for that!).
  • U1B is wrong in the schematic (inverting/non-inverting pins incorrectly labeled) but the layout is correct.

I patched the board and it appears to work. I’m slightly concerned out the output buffer opamp. For some reason it seems to have gain, which causes issues. A new rev is required anyway…