Progress with Hermes-Lite 2.0 beta2

[Dani EA4GPZ]

Hi all,

This weekend I've started building my Hermes-Lite 2.0 beta2. So far,
I've built the power supply, FPGA and Ethernet. All of this seems to
work fine. The Ethernet test works: the board gets an IP via DHCP and
answers ping.

I have some pictures and short videos in my Twitter account (@ea4gpz) in
case someone is interested.

I've made the following modifications, according to the build notes:

* R30, R31 (Magjack LED resistors) 220R instead of 330R. The Magjack
LEDs are pretty dim even with this modification.

* R71, R72, R73, R74 (Orange LED resitors) 1K instead of 330R. The LEDs
are bright, but in my opinion OK.

I've been able to solder all of this with a cheap temperature controlled
iron (and of course, a good tip, solder and flux pen).

73,

Dani.

[Steve Haynal]

Hi Dani,

Great progress on your HL2! Thanks for trying the various resistors. 1K for R71,R72,R73,R74 sounds good to at least save a little current. If 220R doesn't help make the Magjack LEDs brighter, then maybe we should just stick with 330R to save a little current. Those LEDs are on the 2.5V supply and I am worried about drawing too much current from it, especially when an LVDS connection to another unit is used. How hot does the regulator U17 feel in your build?

Thanks also for the tweets. For those without a Twitter account, you can still see Dani's tweets at https://twitter.com/ea4gpz

73,

Steve

KF7O

[Dani EA4GPZ]

El 20/02/17 a las 02:11, Steve Haynal escribió:

> Hi Dani,
>
> Great progress on your HL2! Thanks for trying the various resistors. 1K
> for R71,R72,R73,R74 sounds good to at least save a little current. If
> 220R doesn't help make the Magjack LEDs brighter, then maybe we should
> just stick with 330R to save a little current. Those LEDs are on the
> 2.5V supply and I am worried about drawing too much current from it,
> especially when an LVDS connection to another unit is used. How hot does
> the regulator U17 feel in your build?

Hi Steve,

I haven't tested the Magjack LEDs with 330R, so I can't say if 220R
makes things noticeably better. With 220R they're a little bit dim in my
opinion but still clearly visible. Perhaps with 330R it's much worse.

Although it is difficult to convey this with a camera, I've taken a
picture to show the brightness of the different LEDs:

https://twitter.com/ea4gpz/status/833768860251725828

As you can see from the bloom of the FPGA orange LED, it's much brighter
than the Magjack LEDs. This is with 1K resistors. Perhaps even 2K2 or
something like that will give adequate brightness.

In real life the green Magjack LED also seems brighter than the orange
(which looks more orange than in the picture), perhaps due to the human
eye sensibility to green. It might make sense to use different
resistors, putting a higher value on the green than on the orange.
Taking into account the desire to save current, I think I would be happy
with the green with 220R. The orange needs to be a bit brighter. Viewing
angle is also important. If looked at head on the brightness is OK.

All of this brightness thing depends a lot on personal taste and ambient
lighting (my lab is not very bright, think average home incandescent
lighting). More experimentation and more people's opinions are needed.

After running for over 10 minutes with the LEDs as in the picture (the
green flashing) the 2.5V regulator doesn't get warm. The warmer chip is
the Ethernet PHY (I'm sending a lot of traffic to make the green LED
flash). It is just slightly warmer than my hand, so nothing to worry
about. The 1.2V regulator is also slightly warm, but the rest of the
regulators and FPGA aren't getting much temperature increase.

73,

Dani.

[in3otd]

The MagJack and standalone LEDs will need to have different series resistors also because they are supplied from different voltages, 2.5 V for the former and 3.3 V for the latter. The LEDs may have a drop of more than 2 V (did not measure that yet) so if using the same series resistor the MagJack LEDs get much less current.
Interestingly, even if the datasheet show the led connected to 2.5 V or 3.3 V, the Microchip evaluation board has the LEDs connected to 3.3 V, even if DVDDH is switchable between 2.5 V and 3.3 V (as in the H-Lv2b2), see http://ww1.microchip.com/downloads/en/DeviceDoc/KSZ9031RNX%20Eval-Sckt%20Board%20Rev1_1,%20Sch%20Rev1_2.pdf . Maybe this is acceptable and we could also connect the MagJack LEDs to 3.3 V.

73 de Claudio, IN3OTD / DK1CG

P.S.: thanks for the Digikey shared BOM, Dani. I used that as a base for my order, just added the missing parts you mentioned and reduced some quantities since I already have too many components around, everything was fine.

[Steve Haynal]

Hi Dani,

Thanks for the report Dani. I run at the MagJack LEDs at 2.5V with 330R and they are dim but acceptable to me. We'll see what others say.

I also had seen the schematic that Claudio linked to where the LEDs always use 3.3V. At first I thought that was a mistake so when with Veth. But they are probably using open collector so 3.3V should always be fine. I may switch to 3.3V just to save a little current on the 2.5 supply.

73,

Steve

KF7O

[in3otd]

I also had seen the schematic that Claudio linked to where the LEDs always use 3.3V. At first I thought that was a mistake so when with Veth. But they are probably using open collector so 3.3V should always be fine.

yes, this is what I thought also. The I/O can withstand 3.3 V since VDDH can also be 3.3 V and the usual ESD protection diode between the output and the supply will in practice not be turned on when the LED are connected to 3.3 V and VDDH is 2.5 V since the 0.8 V difference is not enough to have any significant current thru the LED and the diode.
Likely the same if the output is totem-pole also, the LED will have just 0.8 V across it when the output is high and its forward current will be negligible.