define(`TITLE', `Review of network-attached M-Bus master, design version 6')
define(`DATE', `2014-08-04')
define(`CONTENT', `
I got some questions about the design version 6, which I describe <a href="http://a385e-5.de/?p=445" title="http://a385e-5.de/?p=445">here</a> and the modifications I applied to turn it into a network-attached M-Bus master, which I describe <a href="http://a385e-5.de/?p=456" title="here">here</a>.

Here is a kind of a personal review of that design, especially considering the modifications:


<ul>
	<li>I replaced the Arduino Pro Micro by an Teensy 3.1 connected to a WIZNet module.</li>
	<li>I introduced a power supply with three voltages: 34V, 32V and 3.3V.</li>
	<li>3.3V obviously for the MCU, the ethernet module and the 74HC04, 34V for the main analog part, it just replaced the VCC in the schematics.</li>
	<li>Than I removed the four diodes D1-4 and connected 32V directly to the top terminal of R6.</li>
</ul>


While in the design version 6 I had a very clear voltage shift at pin 1 of IC1A of about 100mV for the received logic levels at the bus, this signal was very noisy in the later design. Before replacing the diodes D1-4 by a separate supply voltage, it was nearly unusable and I was not able to communicate reliably using the device.

After introducing the separate 32V supply voltage the signal got better, but still not as good as in design version 6. However, I was able to communicate again.

Nevertheless I ran into another problem: the SNR was going worse for very small and very large bus currents. So, I added a 10k resistor parallel to R7 to drain a minimum bus current of about 3mA and I had to limit the maximum bus current to about 70mA. This is equal to about 70 slave devies. In my setup this is no problem, I use the device just for fun in the basement, in other setups, where actually 250 slaves should be handled, it is a problem. 

I think the major cause for the noise is the stronger, higher clocked, MCU and the WIZnet, which impact noise into the ground lines. (Unfortunately, the Arduino flash is not large enough for my firmware after adding the ethernet code.) And moreover my "breadboard production approach" is certainly a problem. A well layouted PCB with large supply planes leads hopefully to much better results and some more bypass capacitors and ferrit beads would help too.

May be increasing the value of R6 should also be considered. It would increase the voltage shift at the output of a difference amplifier after the S/H device (pin 1 of IC1A).

All in all, considering both the <a href="http://a385e-5.de/?p=445" title="http://a385e-5.de/?p=445">design version 6</a> and the <a href="http://a385e-5.de/?p=456" title="here">network-attached M-Bus master</a>, it becomes clear that it is possible to build a network-attached M-Bus master using the concept.



')