- For the incoming 5v RX signal, suggest you use a single gate chip like the:
74LVC1G08
OR
74LVC1G125/126
Based on your post, I am considering designing another PCB with two sets of 'classic' non inverting level shifters without the 5V pull-ups.
This is the PCB I already made by the way :
Alternatively, I could get the MD8942 switching converter to output 5V (instead of 3v3) and add another linear regulator (AMS1117) to bring it down to 3v3 so that both voltages are available. But this would probably make the PCB too.. packed without installing components on both sides. Which I d like to avoid.
Nice work!
There's still room. I think you'll manage; it's a nice challenge.
The classic level shifter is designed for devices with open drain outputs such as I2C. It's not appropriate in your application, using a serial connection.
Well, before you do that, you should get everything working with a breadboard, including the software.
Also, are you sure your Tx side is working? How do you know? What is the voltage on the AC's Rx input when your circuit is not connected to it?
I agree that normally serial lines would be push/pull. But it's still possible they could work with pullups, at least at some reasonable baud rate, and if this is just for his own AC, not for a product, he might get away with not having to generate 5V just for the Tx line. If, as he says, Tx works now with his open-drain output, with no pullup at all, then there must be one on the AC side that works well enough. Particularly if, as he says, Tx is working now, inverted even.
On the Rx side I suspect the inversion in software isn't working right, and he would be better off with no inversion in either software or hardware. LarryD's LVC125 suggestion would actually be perfect to shift the incoming voltage down to 3.3V, with push/pull. But short of that, the classic shifter with a pullup on the Rx pin could still work. Actually, a pullup resistor with just a reversed schottky diode, or even a 1N4148, could also work.
What's the point, the pull-up serve absolutely no purpose
Well..new pcb already sent for production to China!
Based on two sets of level shifters and local 5v generation.
Sorry but i never use bradbords for trials. Could never get anything working on those! Besides all my designs are SMD (see pic above) and there is no way to get eg a switching converter to work on a breadbord. will post schematic later ..
- You are too fast, was going to say an open drain 74LVC1G07 SOT23-5 would work for both TX and RX level shifting.
@LarryD (In my pic I’m using two UNO’s to test the code and wiring for the project then put on pcb with two standalone 328’s ).
The point really is that you can breadboard bits using other components to test ideas out - so , for example try you level shifter with “ normal” bits then make your PCB with smt .
If you don’t try ideas out , you have take the risk that your min order of pcb’s ($20?) may go in the bin
-
It kinda looked like the purple PCB had two Atmega328 chips and you had two Rev 3 UNOs prototyping . . .
-
Nice LED display plugged into the solderless breadboard.
- I too like to use solderless breadboards to test things, others here not so much.
I suppose your can prototype on bradboard certain small parts of a circuit.
But for a lot of others, its just not possible
For example below is my power supply part for this project.
MD8942 works at 1.1MHz, even if you use some kind of SMD to DIP adapter, its very unlikely that you can get it work on a breadboard. Even on a PCB, unless you closely design part placement and tracks, the chip might still not work properly...
That would ve been a good idea...
Hmm I am thinking to buy some of these single gate chips to have in stock.
Which ones do you recommend besides the above?
Sure you can; you just need to keep the critical part compact. Most of the circuit can be spread around on a breadboard just fine. Will it have quite the same HF noise performance and stability as a proper PCB design? No. Will it work? Sure it will.
Sure you can;
Well.. average row to row capacitance of breadboard is around 2-3pf (rail to rail x10) which translates to about 60k of impedance at the frequency of 1.1MHz. this value is of the same order (or even less) as the resistors that set the operating parameters of the chip.
I didn't say you necessarily plug everything into the breadboard.
You may want to solder certain components directly to each other (dead bug style) or on a small PCB, strip board etc. Then plug that building block into the breadboard and add the less critical parts there.
So this would be like a hybrid approach. That way it can be done, i agree.
Then again it gets complicated and it soon reaches a point where i prefer to make a pcb with the risk of it ending up in the bin!
I find that on average i need to do 2 pcbs before i get one that works as expected.
Maybe you need to brush-up a little on your electronics knowledge