Loading…
Loading…
DIP-16 / SOP-16 · MCP3008
Microchip Technology · DIP-16 / SOP-16 · 8-Channel 10-bit ADC (SPI)
Price on Request
Get a competitive quote within 24 hours. Volume discounts available. Shenzhen stock ships in 1-2 business days.
The MCP3008 is Microchip's eight-channel, 10-bit successive-approximation analog-to-digital converter with an on-chip sample-and-hold and a four-wire SPI-compatible serial interface. It runs from a single 2.7 V to 5.5 V supply, converts at up to 200 ksps at VDD = 5 V and 75 ksps at 2.7 V, and supports SPI clock modes (0,0) and (1,1) with a maximum serial clock of 3.6 MHz at 5 V and 1.35 MHz at 2.7 V. Linearity is guaranteed at ±1 LSB for both DNL and INL with no missing codes over temperature, which is the figure that separates it from unbranded clones.
The conversion is efficient: the CS/SHDN pin doubles as chip select and shutdown control, a conversion needs ten clock cycles of which the first 1.5 cycles are the acquisition window, and the input configuration word selects either eight single-ended channels or four pseudo-differential pairs. Active supply current is 425 µA typical at VDD = VREF = 5 V and 225 µA typical at 2.7 V, while standby current is 5 nA typical and 2 µA maximum — so a battery node can sleep between scans and still measure hundreds of channels per second when awake.
Two architecture points matter in real designs. First, the reference is external: the VREF pin accepts 0.25 V up to VDD, and it defines full scale. Tie VREF to VDD for maximum range and simplicity, or feed a precision 2.5 V reference for better absolute accuracy and drift performance; note that reference current drain is small (100 µA typical) but the reference must be low impedance. Second, the analog input is a switched-capacitor network with about 20 pF of sampling capacitance behind roughly 1 kΩ of switch resistance, so source impedance should be kept low — under a few kilohms for full 200 ksps operation, or add a buffer op-amp. Microchip also warns that very low clock rates (below about 10 kHz) let the sample capacitor leak and degrade linearity at high temperature, so keep the serial clock reasonable even if you convert slowly.
The MCP3008 is the de facto analog expansion part for single-board computers without an ADC, and Microchip supplies it in 16-pin PDIP and SOIC. Family members let you scale: MCP3004 (four channels, same die family), MCP3208 (12-bit, 100 ksps), MCP3301 (differential 13-bit), and MCP3550/MCP3551 (20-bit delta-sigma for low-speed precision). If you need an internal reference, a PGA or I2C instead of SPI, the ADS1015/ADS1115 is the alternative. Sourcing is easy: the part is high volume, priced in the low single-digit dollar range, and stocked in depth in Shenzhen. MOQ is 1 tube or 1 reel for production, or 50 to 100 pcs loose for prototyping, with 1-2 day local dispatch and 4-12 weeks through the Microchip authorized channel. Counterfeit and remarked material does exist in the hobby channel; verify Microchip laser marking and the full part suffix (MCP3008-I/P for PDIP, MCP3008-I/SL for SOIC), and screen a sample for INL and missing codes with a slow ramp input before qualifying a broker lot.
Standardize on the SOIC-16 (MCP3008-I/SL) for volume builds and keep a small PDIP-16 float for prototypes — the SOIC is cheaper per unit, reflows cleanly and avoids socket-related contact noise in fielded units.
Identical architecture and command set, different channel count: the MCP3004 provides four channels in 14-pin packages and the MCP3008 provides eight in 16-pin packages. Firmware that scans channels 0 to 3 works unchanged on either part, so many designs standardize on the MCP3008 to keep one BOM line.
Choose the MCP3008 when you need eight channels at speed from an SPI host and can provide an external reference; it converts at 200 ksps and costs less. Choose the ADS1015 or ADS1115 when you want a built-in PGA and reference, only two I2C wires, or better noise performance at low data rates — the ADS1115 gives 16-bit results at up to 860 SPS.
The input is a switched-capacitor sample-and-hold with about 20 pF behind 1 kΩ of switch resistance, so a 100 kΩ source cannot charge it within the 1.5-clock acquisition window. Buffer the sensor with an op-amp, add a 100 nF reservoir capacitor at the ADC pin, or lengthen acquisition by slowing the clock — while keeping the serial clock above roughly 10 kHz so the sample capacitor does not leak.
Anything from 0.25 V to VDD. Tying VREF to VDD gives full-scale equal to the supply, which is convenient and ratiometric with sensors powered from the same rail. For absolute accuracy use a low-drift 2.048 V or 2.5 V reference and scale in firmware; remember the converter output is proportional to VIN/VREF, so supply noise on a ratiometric setup cancels.
The MCP3008 supports modes (0,0) and (1,1). A conversion frame is a start bit, single-ended/differential select bit, three channel-address bits, then a null bit followed by ten data bits MSB-first — most drivers clock out 24 bits total and mask the low 10. CS must stay low for the whole frame.
Yes, by oversampling and averaging: four samples give an extra half bit, sixteen samples roughly one extra bit, assuming the noise is uncorrelated. For genuine 12-bit or 16-bit results, move to the MCP3208 or MCP3551, which are command-compatible upgrades in the same package family.
Screen with a slow precision ramp or a resistor-divider ladder and check for missing codes, INL beyond ±1 LSB and channel-to-channel gain mismatch. Genuine Microchip parts carry laser marking with the correct suffix (-I/P or -I/SL) and a date code that matches the reel label. Clones often pass a spot check at room temperature but fail at 5 V and 85 °C.
MOQ is 1 tube or 1 reel for production and 50 to 100 pcs loose for prototyping. Shenzhen stock ships in 1-2 business days; Microchip authorized channel replenishment is typically 4-12 weeks depending on allocation.
Submit an RFQ — we reply within 24 hours with stock, pricing, lead time and packaging options.
Request a Quote →