Hermes-Lite 2.0 Build6 Group Buy Shipped

[Steve Haynal]

Hi Group,

All group buy Hermes-Lite 2.0 build6 units have been shipped. Those in the UK have already received their units.

73,

Steve

KF7O

 

[Steve Haynal]

Hi Group,

Here is an update on the recent build6 of 25 units. They have all been shipped and 23 out of 25 have arrived. The remaining two units were shipped air mail to India and Australia and require extra time. Although there hasn't been as much information posted to this Google Group, there is a private e-mail thread among all the group buy members. Several have reported success with RX. I haven't heard any TX reports yet as that requires some finishing steps. No one has complained or brought up any serious problems yet. :)

Overall, I've been happy with my build6 unit. The changes made to reduce the AD9866 temperature (remove manual soldering hole for thermal pad, unblocked paths for heat on inner layers) did not seem to make much difference. When we thought we had build problems, Elecrow did remove one of the AD9866 and verified that it was properly soldered to the thermal pad, so I doubt if that is an issue. Maybe there will be some ideas for build7 to improve this. Build6 users may want to consider a small heat sink, although I haven't bothered with one and just let it run hot. We are not exceeding any specified limits.

73,

Steve

KF7O

[pascal.v...@gmail.com]

Hello group,

Happy to say the HL2b6 is running well with the same full power (6W without LPF) on all the bands. Good sensibility ( much better than my 1.22 protoype)

Thanks for the team for the nice result. I have now to populate the N2ADR LPF before to close the box.

Pascal F6EHP

[Steve Haynal]

Hi Pascal,

Great to hear that TX is working for you, and that this HL has the expected receive sensitivity.

73,

Steve

KF7O

[pascal.v...@gmail.com]

Hi Steve,
The N2ADR LPF is now online and seems working well with the HL2.
I wonder why the full power on a whistle (2.5/3W) is only HALF of the the full power on a spot tune (5/6W) ?
The same software (quisk) is used with the HL1,22 on the same sbc and gives the same power level on a whistle and on a tune.
73.
Pascal

[Steve Haynal]

Hi Pascal,

How are you measuring power in both cases? If the LPF is in place, do you have the band selection correct so that the filter is not attenuating the signal? There is a rare bug we've been tracking where the HL2 sometimes powers up in a state where it is not able to achieve full power. If you power cycle the HL2, does the low power output still persist? The WSJT-X software tune function provides a more repeatable "whistle." I see the same power output with this tune as I do with Quisk spot. Can you try this?

73,

Steve

KF7O

[pascal.v...@gmail.com]

Hi Steve,
Measuring is done with a wattmeter and with the PA current given by the N2ADR LPF : 1050mA with a spot, and not more than 800mA with a whistle. ( Selection of bands is ok and I get the same power ratio without the LPF). Power cycling the HL2 doesn't change anything. Moving from one sbc to another doesn't change.To try WSJT-X tune, I need to allow the 5W PA in PowerSDR. How to do that ? I got it  once selecting "enable tuner" in the parameters, but now it stays off .
73
Pascal

[Steve Haynal]

Hi Pascal,

I suspect something is wrong with your audio setup. The difference between spot and an audio signal in is just that spot sends a constant IQ value. All the processing in the HL is the same for spot and audio. Since you are seeing proper output for spot, I don't think there is anything wrong with your HL2, but something is wrong with your audio, microphone or Quisk configuration. Please provide more details regarding which Quisk version you are using, which OS, which soundcard (alsa, pulse, other), and exact steps to reproduce your problem.

Please try SparkSDR and report if you see the same problem.

You don't need PowerSDR to try WSJT-X tune. Newer versions of Quisk include "Virtual Cables" already setup at least on Linux to connect to WSJT-X. Below is a screenshot of tune in WSJT-X. Note that FDX is on so I can monitor the output signal, and the waterfall display is selected. Power output is 5W. What do you see when you try the same?

Below is a picture from the TX config of Quisk. Note that when transmitting tune from WSJTX, the levels are adjusted so that the TX audio peaks at -1 dB. What do you see?

73,

Steve

KF7O

On Sunday, May 13, 2018 at 2:16:59 AM UTC-7, pascal.v...@gmail.com wrote:

Hi Steve,

[pascal.v...@gmail.com]

Hi Steve
All tests are made with the Quisk 4.1.15 on linux or windows10 sbc. Using two different headset/mic.
Spark is not yet working here. I try to solve that with the help of Alan.

I have connected WSJT-X on quisk on my windows sbc ( unable to install WSJT on linux for the moment and not sure I could do on an arm cpu ) , but I find strange that I must select the usb mode to get an output power with the wsjt tune, and get no power with the DGT-U mode as you show on the photo (digital output set to 100%) ? Anyway, the output power stays under the spot power as with any mic.(-1dB audio level)

[Alan Hopper]

Hi Steve,

does the build 6 report itself differently to beta3 in the discovery?

73 Alan M0NNB

[Steve Haynal]

Hi Pascal,

For DGT-U do you have the following set:

1. Slider "Digital Tx level" in config screen set to 100%. 

2. Entry box "Digital Tx power %" in your Radio Hardware screen set to 100%.

3. Digital Input and Output audio devices in the Radio/Sound screen (not top level sound config, but under a radio configuration) set to valid audio devices.

4. Tx Level set to the full 100% for all bands in the Radio/Band screen.

If the above is true, you should be able to see WSJT-X tune output. All this is documented in the YouTube video:  https://github.com/softerhardware/Hermes-Lite2/wiki/Hermes-Lite-2.0-Getting-Started#quisk-setup

Can you verify that if you make these change then you see full output with DGT-U? Note for WSJT-X you may also have to adjust the output up with the slider on the bottom right of the WSJT-X screen.

If I do the same as you and route the WSJT-X output to the "Microphone Input" and use USB, then I also see reduced power 3.8W. I could not find any changes to achieve full power with USB. But I do see full power with DGT-U. This makes me think there is a problem with Quisk (additional filtering/processing for USB vs DGT-U) that is attenuating the same signal from WSJT-X, or some other Quisk assumption for the Microphone input that we are not following. I hope Jim can answer this.

73,

Steve

KF7O

[Steve Haynal]

Hi Alan,

No. Still the same board ID and code revision.

73,

Steve

KF7O

[pascal.v...@gmail.com]

Hi Steve,
Yes, I was wrong on point 3 where I didn't put the virtual cable in and out on the digital lines.
Now the tune from WSJT gives the same power as spot.
So, we see the same issue with the ssb mode. But I must say I don't see the issue with the same quisk release and the previous HL1.22.
Thanks for your help.

[James Ahlstrom]

Hello Pascal and Steve,

For my HL2, I see 5.0 watts on Spot and 3.6 watts on a whistle as measured by the Quisk power meter. For my HL1 I used my spectrum analyzer and measured 19.31 dBm for Spot and about 17.9 dBm for a whistle. The whistle signal is noisy, perhaps because I am getting wind noise by whistling into the mic. But 19.3 - 17.9 is 1.4 dB, and 10*log10(5.0/3.6) is 1.4 dB so I think the HL2 and HL1 are acting the same, and the difference is in Quisk. Remember that for a digital radio, the issue is peak power because we can not allow the DAC to clip. So to correctly measure PEP be sure you have a peak power meter, not an average meter.

The audio processing for SSB is much different from Spot or digital.  For Spot, quisk sends a constant value at the peak DAC level. For digital, quisk assumes the signal is clean and at peak DAC level, so it uses minimal processing and a bit of ALC. So Spot, and to a lesser extent digital, can both get close to the DAC maximum.

For SSB, the mic signal is filtered to 300 to 2800 hertz, then amplified to a constant level and clipped, filtered again, and then ALC is applied. Without this aggressive clipping and filtering, the average power for 5 watts PEP would be less than one watt. For a clean sin wave into the mic input, you could expect 5 watts PEP. But the system is designed for speech. If I get too close to the maximum DAC level, I risk clipping on a sudden loud noise like the letter "P". There is a bit of black magic to this speech processing, and it lacks the clean predictability of digital. I could perhaps increase the SSB level a bit closer to DAC maximum, but remember that one S unit is 6 dB, so one S unit below 5 watts PEP is 1.26 watts. I think the difference in actual PEP is less than the 1.4 dB measured and that this difference does not matter in real life.

Jim

N2ADR

[pascal.v...@gmail.com]

Thanks Jim for this clear explanation. I certainly overdrive my HL1.22 PA to get so much power on a whistle.
My spark is now working , and for your information, a whistle gives the full output power.
73, Pascal

[Amogh Desai]

Hi Guys,

I received my HL2 board yesterday.  Basic receive testing is successful. Below is a screenshot of a local AM station with a few centimeters of wire as antenna.

Thanks Steve for all the efforts and everyone else who helped shape this beautiful radio !

Regards,

Amogh

VU3DES

[Steve Haynal]

Hi Amogh,

Thanks for the update! I'm happy to hear your HL2 arrived safely.

73,

Steve

KF7O

[John Marvin]

Hi All,

I also got a Build 6 HL2, but only today was able to get it connected to
my antenna. It seems to be working fine. Currently running WSPR on 20m,
30m and 40m using all 3 receivers and getting about 350 contacts / hour.

The A/D doesn't seem to be too hot, I can hold my finger on it without
it being uncomfortably hot. Don't know how much hotter it would get with
simultaneous transmit, but if it's not much hotter than this I don't see
a need for a heatsink.

Which brings up my next question. I'm probably not going to put this in
the suggested case, and am wondering what ideas, if any, others have had
for heat dissipation for the 5 watt LDMOS output devices. I imagine
those things get pretty hot during transmit and would die a quick death
without adequate heat sinking.

73,
John
AC0ZG

[Anthony Loveday]

Hi John,

I have used four (2x2) TO220 heatsinks clamped to the edge of the board using a bolt through the mounting hole to dissipate heat in the devices.

You'll note I don't have the TR relay installed - I fitted J14 and will be using an off board changeover and filters.

73,

Tony

VK4KRC

--
You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[John Marvin]

Tony,

Thanks for the pictures.  I'm a little surprised your heat sinks aren't clamped to the top of the devices themselves. Do you believe you are getting adequate heat dispersion with this setup? Have you made any attempt to measure the temperature at the top of the device (either temp sensor or infrared detector)?  I was thinking that perhaps mounting something similar to you but using the single hole between the devices instead. I'm not sure if a single TO220 heat sink on each side (along with some thermal compound applied to the tops of the LDMOS parts) would be able to properly cover both. But I'm still considering options and am glad you shared this.

I am also not planning to install a TR relay either.  I am going to use the QSK board (available separately) from the HardRock 50 amp. It's a little pricey ($49) but my whole amp chain does pin-diode switching so I prefer to keep the whole chain that way.

I don't plan to use a filter board.  As long as I control the output levels, I'm fairly sure the filters in my amps will adequately filter the harmonics.  In the rare case I might operate without the amps , I have an external dedicated filter for 40m that I can connect, which of course limits me to operation on 40m in that configuration. Normally, even if I am running "QRP", I adjust the drive level such that my KXPA100 amp is putting out 5 watts, so that I am still using the filter bank within the amp, although that kills the amp efficiency.

73,
John
AC0ZG

[Anthony Loveday]

Hi John,

Yes, I agree - the transfer will be better with the heatsinks clamped to the top of the LDMOS devices. I just had them there to try something similar to the rails in the enclosure while I set up the board to transmit. The temperature rise was acceptable during that exercise but I'll send some temperature readings once I relocate to them on to the transistors. I'm presently just receiving until I have the filters finalised.

73,

Tony.

VK4KRC

[Heikki Ahola]

John,

My solution to the PA cooling is described in the thread  "HL2 alive in OH country !". I used an aluminium block under the board and a copper plate above, clamped together and to the chassis (a large instrument cabined salvaged). In my case, he copper plate (1.5 mm thickness) is firmly pressed to the metallic cooling pins of the two PA  transistors. See the pictures provided  in the thread. 

73 de Heikki (OH2LZI)

[Steve Haynal]

Hi John,

I think you will find that the PA dissipates less heat than you are expecting. We run the PA at ~8V so there is less heat generated. The current solution is adequate and only dissipates heat along the thin edge rails. Places like Home Depot sell some copper strips, maybe a few millimeters thick. I always thought stacking a few of those on the bottom exposed area under the LDMOS devices would work well.

73,

Steve

KF7O

[Dieter Kedrowitsch]

I recently built a Hermes Lite 2.0 which uses the "build 6" PCB design, and I've not supplied any additional sources of LDMOS cooling besides what's designed into the PCB itself.  I run it's PA flat-out for long periods of time without any heat issues whatsoever so this point, supplemental cooling is probably unnecessary except perhaps in more extreme cases.

Dieter 

W3DRK

[Alan Hopper]

Hi Dieter, group,

was that in a standard case or bare board?  I'd be surprised if it is happy without anything.  The standard case seems ok, I Think it works slightly better with the board jacked up to the top of the slot rather than bolted down to the bottom as the ends of the bottom part of the slot are distorted by the end plate threads (in my one anyway, others may be different). For testing I had an aluminium block the size of the rectangle of exposed pcb under the mosfets and bolted it to a oversize heatsink and it stayed very cool indicating that the rectangular pad is well connected to the mosfets thermally and is probably the simplest was to get heat out if not using the standard case.

73 Alan M0NNB

[Dieter Kedrowitsch]

I've run it bare and in the standard case without trouble.  Also, I opted for a black powder coated case--so no direct contact to the aluminum either.

Dieter