But... with a shift register, you can save about 1000 wires and eliminate the need for CP/OE signals.
Schematic or it didn't happen.
I see the "wires" being the connection of 128 points from the mystery box to the circuit board with the CP/OE being traces on the PCB so I will need you to explain where the 1000 wires come from.
I mean the chip connections to the parallel buses. Of course, if they’re PCB traces rather than hand wiring, then it’s about 1000 tracks (64 chips considering the whole system, with 8 signals each).
- Some ideas:
- If multiplex outputs are used, the SPI control of a single MAX7221 lets you have
64 outputs. - The simple 74HC595 can be filled in1μs using SPI.
- The TPIC6B595 gives high drive capability.
- MCP23017 is 16-bits, input or output.
- PCF8575 i2c 16-bit I/O expander
- If multiplex outputs are used, the SPI control of a single MAX7221 lets you have
There are many good options.
Consider that to minimise your processing and comms, you only need to transmit the bits that have changed state between samples…
This could dramatically reduce the comms payload if only a few bits change.
EDIT:
You could add a cycle that when requested, or on every tenth sample, to refresh the whole array.
Very, very interesting. I'll do it this way: since there's a long way to go between saying and doing (as they say here). In other words, once the system is built, some unexpected problems can always arise. So, once the first little project is finished, I'll test it, and after a certain period of time, if I want to make some changes, I can try the other various options to compare them and see if they can prove to be a step forward.
Thank you all. I'm happy to have seen your great availability and speed. Now it's my turn to work hard.
P.S.: I don't understand exactly what the paid consulting section is on the forum. I can't find the page...
Very interesting. Must try....
Thanx
Just for GENERAL KNOWLEDGE. This schematic I posted belonged to a three-digit voltmeter/ammeter with a red LED display. The three-digit voltage reading (ranging from 00.0 to 99.9 V (A)) was displayed remotely via RS232. This circuit simply transmitted the BCD logic states of the LED display drivers, which were very common CMOS devices of the time.
Interesting, right? Also because if the voltages were converted to RS485, the numerical data could be transmitted over a much greater distance.
In fact, I assume that when 56kHz modems with RS232 serial ports came out, perhaps the data could even be transmitted over the internet. Fantastic, I'm talking about years ago now.
What I find incredibly strange is that in all these years (I'm talking about the 1990s and later) I've never found anything for sale that did this before complex microcontrollers that only communicated via USB or LAN. USB obviously has the disadvantage of requiring a PC always running on the remote site (and LAN was much later). Let me explain: perhaps I'm the one who wasn't informed, but essentially I'm talking about a remote measurement system that wasn't via LAN or USB and displayed voltages/currents from special meters on a physical display (and not a PC) so that remote readings could be obtained via RS232 or RS485.
I've searched for years but haven't found much more than the schematic I posted, which is from the late 1980s! Is that possible?
Can anyone enlighten me and tell me what similar stuff was for sale? Also, could you provide some links, photos, and/or part numbers/models?
P.S.: It's true: you never stop learning....
My brother worked in power stations, from what he told me all the signalling and monitoring for the control systems was via twisted pair cables.
Very interesting. I'd be curious to study the circuit diagrams of those devices and how many digits they could transmit remotely. I'm passionate about old technology and schematics with CMOS, TTL, DTL, and RTL...
This seems very good, except I forgot to check if it's one-way because I need duplex (half or full). I couldn't use two (one in one direction and one in the other) because then I'd need two separate RS-232 lines, or am I wrong?
Is the arrangement identical on both ends? From your description, I thought it was more a master/slave arrangement. Is there a display of the opposite side on both sides or is the data mirror just on one side.
Maybe a block drawing of the setup would help explain it.
The RS-232 is duplex but the way I envisioned it, there was just a data transmitter on one end and the dat mirror on the other.
Forget about RS232, use RS422
You will need 4 twisted pairs, so an Ethernet cable can be used
Two pairs for full duplex data and two pairs for full duplex sync.
Speed up to 50Mbits/sec.
While the latch approach I described is intended to preserve the old school nature of the circuitry as well as a convenient way to take a snapshot of the instantaneous data states there are more ways to do it with more modern parts. I/O expanders like the MCP23S17 might be more convenient to use. Not sure how to command multiple 23S17s to take a snapshot but other chips might work better.
Maybe other helpers will chime in. I'm pretty busy today.
Yes, yes. And it all works in two ways, in the sense that on the one hand you receive and transmit, and on the other also.
Speed isn't the issue. Even a dual RS232 can be used across two pairs...
True, so what is the problem?
the problem is that the existing line must be used which is 2 wires + screen