Take me home
SCi:
CDi3.FTe Submersible, Universal Gas-Field Controller
Keller Digital Pressure Transducers/Transmitters
Burkert Solenoid Modifications
PDL/CDi History: Pembina Lake Erie Datalogger
LD Series Pressure Transmitters
Projects:
CDi4.P8 Flowmeter, Keller Communicator
CDi4
LD Series Pressure Transmitters
FlyportPRO
Internet of Things - BME
Internet Enabled, Tomato Garden Watering Thing
ZDL Potential Boards
P5 Evolution Project
CDi4, ICISS2 **** PRELIMINARY PROPOSAL INFORMATION ****

The ICISS1/CDi3 used the DIP version of the Zilog Z8S180, the CPU, it was limited to a maximum clock rate of 10MHz. There was a 20MHz maximum DIP version manufactured some time ago, we only managed to buy one but that's now irrelevant because all DIP packages are obsolete. Top family clock rate is 33MHz and one can buy 33MHz chips in the marketplace, but only in QFP and/or PLCC packages.

The QFP package (local picture) is a Surface-Mounted Device, SMD, so it must be soldered directly to the PCB traces, the PLCC package (local picture) of the chip is also an SMD but this one can optionally be socketed in which case we are likely to have a problem because the socket's height is over 1/4" and this might make it difficult to fit the flipped memory board right over it. We will most likely select the QFP version, the whole assembly is moving from through-hole to SMT.

CDi4 Flowmeter/Keller Communicator

The 2000 ICISS had a 6.144MHz crystal in it but its clock rate was only half of that due to a programmable divide by two flip-flop inside the chip. The recent modifications and upgrades brought about the usage of a 7.3MHz crystal and this resulted in a unit with both low and high speeds, namely, 7.3MHz or 3.65MHz. The 7.3MHz mode results in a maximum baud rate of 115.2K, if we were to increase the clock rate four-fold we would also increase the baud rate four times to about 460.8K bits per second, roughly 46K bytes per second, so a binary download of all storage memory would take 25 seconds or so. If we double the storage memory to 2 MB we can still dump all that in binary mode in less than a minute. Some serious testing to be done here, how much do we increase the clock rate, do we double it, triple it, quadruple it, and does the increase affect reliability, and if so, how? The 20MHz version of the DIP mentioned above was used in a test unit, operation at 14.6MHz has been verified with no issues.

A minimum of 5 PCBs must be designed no matter what, a Main Board PCB (1), two Memory Board PCBs (2) already done see below, a Power/Charging Board PCB (1) and a Keyboard PCB (1). It is likely that a sixth PCB will be needed to house the analogue circuitry with the ADC and the microcontroller, see the ADC section below.

So the job description is:

  1. Double, triple, or quadruple the clock rate from 7.3MHz to 14.6MHz, 21.9MHz, or 29.2MHz
  2. Double the storage memory from 1MB to 2 MB
  3. Potentially double the EEPROM size from 64K x 8 to 128K x 8
  4. Get rid of the lead-acid battery/charging, replace it with NiMH which will instantly more than double capacity
  5. Make the move from ±15-bits ADC to ±23-bits ADC and offload the job, min/max, qsort, etc. to a microcontroller, automate calibration
  6. Get rid of that ancient keyboard encoder already!
  7. Get rid of that DB9 serial connector at unit front, include USB to serial circuitry within unit, outside only needs USB connector
Keyboard
The 74C922 keyboard encoder has been obsolete for years and needs to be replaced. A suitable replacement is the PDN1188 (external pdf link), the chip can decode an 8x8 key matrix so it can differentiate up to 64 keys. A suitable replacement for the Grayhill keys has not been found so we're stuck with the pricey but reliable Grayhills. Obviously we'll need a new keyboard PCB designed.
Display
We're keeping the 4x20 character display with backlight, however, the new display doesn't have to have a parallel interface as other serial options exist such as SPI or I2C. A serial interface to the display cuts down the number of lines from 16 to about 6.
Battery/Charging
Lead-acid batteries are history, we've replaced them in the lake-related gadgetry in 2010, and so have well over 10 years of field experience in using the NiMH chemistry. The move to Nickel Metal Hydride batteries will more than double capacity from the current 1.1Ah to about 2.6Ah. So not much to do on the electrical design phase, we'll need a new battery/charging PCB, obviously. There are multiple manufacturers and we've been using the "D" size AccuLoop ones successfully for a decade, in the CDi4/ICISS2 we'll be using 5 of the "AA" size like these (external link).
Memory/EEPROM
The new PCBs are physically the same size as before, but have double the storage RAM from 1MB to 2MB and can potentially double the program chip size from 64K x 8 to 128K x 8.Why two PCBs? Just in case the DIP version of EEPROMS disappear (left picture). Also worthy of mention here is the design team has been moving to the Open Source KiCad Design Suite (external link) for a couple of years now so even though these two PCBs are already designed, they are not designed in KiCAd, meaning, they'll probably need to be redone, copied, to a KiCad design.
CDi4 Memory Module 1
  • Board status: designed June 2016, not built yet.
  • Doubles the ICISS1/CDi3 storage Static RAM, SRAM, to 2 Mbytes.
  • Accommodates 64k x 8 and 128k x 8 EEPROMS in 32-pin DIP packages.
  • Board pdf: layers combined, board pdf: layers on separate pages.
CDi4 Memory Module 2
ADC
The current ±15-bit MAX135 ADC is the slowest peripheral on board, it has a parallel interface, it needs to be replaced with a serial ±24-bit ADC like the NAU7802KGI from Nuvoton (external ink). The new ADC is not likely to be controlled by the CPU directly, rather the CPU will command a microcontroller to fetch a reading or a reading set like a min/max. Lots of work to do here.
RTC
The current DS1386 RTC has been obsolete since 2008, it has been replaced by the DS1554 (external link) which is a completely different animal with a totally new, and incompatible to the DS1386, register set. This has already been implemented in code, the user will never know which RTC is implemented in the unit, in other words, no new operational instructions to the user, the implementation of the new RTC is 100% transparent from the user's point of view.
Communications
MRU-AF41-1000Very small USB to serial converters can now be purchased, we should include one inside the unit, so we only need a suitable connector and a standard USB patch cord. Example connector to replace the DB9 serial connector in the picture on the left, click for larger image.
Contact:
SCi.RS Cloud (Mirror of singlechips.com)SingleChips Greece Cloud (Mirror of SingleChips Greece)NuovoPicus CanadaXYZ CanadaSingleChips GreeceSCi.RSCensors!!!Censors!!!Censors!!!
Copyright © 1991-2024 AdamG Last Update: Monday, June 12, 2023 2:48:39 PM