Thermometer with DS18B20 sensor and two-digit LED display

AT89C2051- I started my uC adventure with it 20 years ago. Did you write the procedure for handling 1Wire and communicating with DS18B20 yourself? How do you limit the current of the diodes in the display as I didn't notice any limiting resistors? Greetings

Andrzej_Tomaszewski wrote:

The procedure for handling 1Wire and communicating with the DS18B20 you wrote yourself?

Yes, these are my battle-tested procedures, here used in the trivial case of a single sensor on a ~20cm wire. I converted their delays to a 12 MHz clock. So far the smallest pull-up I've had was 1.5k, here it happily runs at 1k, so the ladder from the display takes care of the 1-wire too.

Andrzej_Tomaszewski wrote:

How do you limit the diode current in the display because I didn't notice any limiting resistors?

The 1 kΩ resistor ladder limits the diode current in the display. This is a display with a common cathode. The current flows from plus through the 1k resistor to the anode of the segment, from the cathode it flows into either port P3.4 or P3.5 and through the CPU to ground. If I want to extinguish a segment connected to P1.X, I put a zero there and then the current flows immediately through the processor to GND instead of to the LED. That is, the current consumption is slightly higher when the segments are not lit than when they are lit. I took this clever way of simplifying the circuit from the kit https://www.elektroda.pl/rtvforum/topic3494198.html

Well yes, if the WK display is ok but the control method is a bit unusual, not very energy efficient. Regards

Added after 2 [minutes]:

Sam Sung wrote:

That is, the current consumption is slightly higher when the segments are not lit than when they are lit. I took this clever way of simplifying the circuit from the putty https://www.elektroda.pl/rtvforum/topic3494198.html

And haven't you thought about a resistor ladder with individual leads or just individual resistors?

Andrzej_Tomaszewski wrote:

And you didn't think about a ladder with resistors with individual leads

I didn't know there were such.

Andrzej_Tomaszewski wrote:

or just about individual resistors?

This kind of circuit would not fit on such a small board.

Sam Sung wrote:

A circuit like that would not fit on such a small board.

Just out of curiosity - what was dictated by this board size? It seems that the enclosure allowed the use of a slightly larger board.

Years ago I used assembler, then fell in love with C++, I remember from assembler that it was like ploughing acres of field with a hand plough.

Respect and admiration, especially for your original solution and perseverance.

If it wasn't for that trick with the display control, you wouldn't be without a few transistors or some ULN28.... for classic control.
The AT89C2051 itself has probably not a lot of capacity, surely it can't light up 8 segments and power it from the proc ports.

The following shows that the output in the low state can accept 20mA while in the high state the current flowing out is negligible. I wonder what was dictated by the choice of this and not another uC.

Sam Sung wrote:

a clever way to simplify the circuit

Way from PCF8574 note

Andrzej_Tomaszewski wrote:

Wondering what was dictated by the choice of this and not another uC.

The choice of board and prock was mainly dictated by the fact that I had one, but I also wanted to see if I could do such a small chip with such an old proc. The port load is high, the ports have a strong zero, so newer procs have no advantage here. Well, unless the 1-wire could work stably without a quartz - then the AT89C2051 would lose out with the lack of an internal RC oscillator.

ex-or wrote:

Sam Sung wrote:

a clever way to simplify the circuit

Way from the PCF8574 note

And from which one? I don't see such a diagram in SCPS068J.

Sam Sung wrote:

The choice of board and prock was mainly dictated by the fact that I had one, but I also wanted to see if it would be possible to make such a small chip with such an old proc.

The microprocessor used is very cool, for many one of the first in an electronic career, due to BASCOM. The relatively small amount of ROM is enough for small projects and to practice assembler, because the program does not grow too much.
the 1-WIRE pulls it off without a hitch, even in BASCOM it was possible to do sensible things.

In the course of developing the program, how many times did you insert the 2051 into the socket? I suspect that the lack of ISP becomes a nuisance here, especially when debugging. I've used it a few times in projects and I remember the legs stripping down to bare copper from constant programming (I was using a precision cradle).

acctr wrote:

1-WIRE pulls it off without a hitch, even in BASCOM it was possible to do sensible things.

Bascom was not as bad as it was painted. Here an example of a thermometer on an ATTiny2313 with a DS18B20.
http://mirley.net/minitermometr_uniwersalny.html
This project is also on the electrode as far as I remember.

Sam Sung wrote:

I didn't notice with the naked eye the effect of such multiplexing attachment on the brightness of the display.

The author @mirley has written up the sensor comms in an interesting way. Worth a look. I'll just add that on a similar code I made a temperature controller on an Atmega8 for an etching machine. As a curiosity I will add that I tried to do it in Arduino IDE. (MiniCore/Atmega8 board). Unfortunately, the library alone to handle the DS18B20 was taking up more than 80% of the flash of this probe. In Bascom I did it with a lot of slack

efi222 wrote:

Bascom wasn't so bad

And it still isn't

acctr wrote:

In the course of creating the program, how many times did you put the 2051 into the cradle?

Many times. Well it didn't want to work the first time

efi222 wrote:

Bascom was not as bad as they paint it. Here an example of a thermometer on ATTiny2313 with DS18B20.
http://mirley.net/minitermometr_uniwersalny.html
This project is also on the electrode as far as I remember.

Hmm, I've done error handling though, and I don't keep flying by no matter what happens on the bus
Well, and I put the microcontroller to sleep - the interrupt wakes it up. And it's not possible to do sleep in Bascom?
I don't really understand why I need to divide the same number modulo 10 again every 4 ms by chunking the same digits and converting them into the same segment patterns when the number changes every second. The sentence "the Mux=4 state is the phase waiting for the temperature measurement to finish quietly" is probably some nonsense (?) To my mind the Mux=4 state only reduces the brightness of the display from 3/4 to 3/5. In my case I did not introduce a dead multiplexing phase because this 1k ladder limits the current sufficiently. With dimming the brightness is OK.
In 2011 I also made a thermometer that works a bit like that mini thermometer, i.e. for the duration of all communication with the 1-wire the display was left blank. That's how I know that this causes a rather annoying blinking. Meanwhile, nothing bad happened when here I let multiplexing run in an interrupt, which I pause for the duration of a single operation on 1-wire. There is no flickering (unless of a different kind, when the temp is at the limit and changes by 1 degree every measurement), the display has not burned up.
Still, these multiplexing phases are written out in bascom in a kind of vascular way, like switch-case in C, as if they were some kind of special case, instead of an elegant mask with a 1 or thereabouts zero turning on a single digit, going around in circles.... well no, this code did not win any sympathy for bascom

Such a curiosity: in the DS18B20 note there is a table with temp measurement times at different resolutions:

And a long time ago the sensors worked like this. Instead, the ones I bought recently on AliExpress seem to measure just as fast, no matter how many bits of resolution you set. They don't even try to pretend to be original? Well, but in parasitic power mode you can't tell.

Sam Sung wrote:

Well, I put the microcontroller to sleep - an interrupt wakes it up. And in Bascom it is not possible to do sleep?

In Bascom, one short command and the uC sleeps, error detection on the 1Wire bus - it's just testing the state of a dedicated variable, processing data from DS18B20 and converting it to degC with simultaneous dynamic display - a fairy tale, in addition to simultaneous operation of the dynamic display and menu operation in which, among others, I change the brightness of the lights - no problem.

Best regards

saly wrote:

Years ago I used assembler, then I fell in love with C++, I remember with assembler it was like ploughing acres of field with a hand plough.

Tum no less sometimes it's nice to insert an insert in asm .....

Bravo for the assembler. Pity not at2313 .

astony wrote:

Bravo for the assembler. Too bad not at2313 .

When you write identical functionality in assembler on ATtiny2313, we'll compare the conciseness of 8051 vs AVR machine code
It seems that the AVR must lose by a landslide because:
- it's RISC
- there are no single-byte commands
- there is no hardware division (by 10)
- even the GPIO has to be set up

Sam Sung wrote:

must lose by a whisker

But in which category? Because certainly not in performance.

Sam Sung wrote:

Andrzej_Tomaszewski wrote:

or just about single resistors?

That kind of circuit wouldn't fit on such a small board.

Your board would fit another twenty resistors without any problem.
Using SMD resistors instead of chip resistors. SMD resistors of size 0805 and 0603 fit best on such universal boards. The use of SMD allows, among other things, the use of areas covered on the other side, e.g. the base or the display. If the boards have metallized holes, wire connections can be made both on the component side and on the other side. And just using these resistors in such a context, in my opinion, is quite a lot more convenient than the through-hole ones.

Here, for example, is a typical universal board with a 2.54mm grid and some resistors of both the 0805 and 0603 dimensions, where you can see how much space is left on the eyelet for the wire: