Hexadecimal displays

[gbm]

Most of contemporary CPUs have built-in boot loader, so all you need for a "bootstrap" is USB-UART logic level converter.
And relying on CMOS output resistance is not a good design practice...

[Chromatix]

1024MAK wrote:

Who says old school logic only runs at 5V?

A PIC16C84-04/P microcontroller supplying counting pulses to a SN74LS393 counter, which in turn is supplying binary data to a MC14495P display driver. This then is driving a red seven segment display. The whole lot is being powered from two AA cells, the current battery voltage is 2.93V as shown on the cheap multimeter
The MC14495P and the common cathode red seven segment display were obtained from China via eBay.

That's just about perfect! Very good to know that that chip works so well outside its official spec.

[BitWise]

Chromatix wrote:

1024MAK wrote:

Who says old school logic only runs at 5V?

A PIC16C84-04/P microcontroller supplying counting pulses to a SN74LS393 counter, which in turn is supplying binary data to a MC14495P display driver. This then is driving a red seven segment display. The whole lot is being powered from two AA cells, the current battery voltage is 2.93V as shown on the cheap multimeter
The MC14495P and the common cathode red seven segment display were obtained from China via eBay.

That's just about perfect! Very good to know that that chip works so well outside its official spec.

Its well within its spec at that voltage but it might limit the maximum operating speed.

Code: Select all

Wide operating voltage range:
- Commercial: 2.0V to 6.0V
- Industrial: 2.0V to 6.0V

The datasheet suggests a top speed of around 4MHz at 3V3 compared to 5MHz at 5V.

[Chromatix]

I meant the MC14495P, not the PIC or the 74HC.

[BillO]

I just got my MC14495P1 chips and can verify they do work at 3.3V. However, my guess is they are gonna be real slow that that voltage. @ 5V the typical LE pulse width is ~200ns up from 80 ns @10V. I haven't done any tests yet, but my guess is you probably won't be able to drive them directly from a bus running at much over 1 Mhz @3.3V ... if that.

[Chromatix]

That's fine, since if I cycle through 6 or 8 digits at 32kHz (ie. drive them off the RTC oscillator), the digits will still be refreshed several thousand times a second. And that's if I multiplex all the digits through one driver, rather than providing a driver per digit - I'll need to check the cost-complexity tradeoff.

[BillO]

I got 10 for about $18 from eBay seller adeleparts2010 including the shipping.

[drogon]

Just seen this:

https://www.youtube.com/watch?v=7NClJQR-1c4

which may be the right thing here?

-Gordon

[BillO]

I know this thread is nearly a year old, but ...

Today I was going through an unsorted parts bin where I place things I intend to use some day

And I came across little baggie containing some Maxim ICM7218. Seems like a very capable display driver.

Looks like there are/were lots of people making it too.

https://www.maximintegrated.com/en/prod ... M7218.html

[cjs]

Chromatix wrote:

...thinking, at least theoretically, about how to built voting machines that could actually be trustworthy. That means they have to be...auditable from the hardware level upwards. A 6502-based system could be a good basis for that, since everything is big and crude enough to examine on an airport-grade X-ray machine and the source code would be small enough to examine by hand, but the number of programmable devices embedded in it would have to be strictly minimised. A GAL chip or a PIC stuck in the middle of the bootstrap process would be a giant red flag.

This is indeed a thing. See Princeton's Vintage Verification project for use of 6502-based machines for nuclear arms control verification, for exactly the kinds of reasons you're talking about. (It's important that the machines being used can be shown to the inspectors to be working properly, and to the inspectees not to be revealing nuclear signature data outside of the immediate inspection process.)

[Chromatix]

Very interesting indeed. I wonder if they'd be interested in my approach to verifying the 6502 type in use - though they'd probably want a more thorough verification than I'm aiming at with my 6502 Fake Finder.