Hermes-Lite Group Buy Build9 Discussion

[Steve Haynal]

Hi Group,

It is time to start working on the next group buy, build9. Here is my list. Any feedback is appreciated. Many of these items are tracked on the github issues page.

** Makerfabs will price at 25 units again. The group buy will be open for 4 weeks. I am trying to move more overall responsibility to Makerfabs, but with every group buy there seems to be some new surprise...

** Makerfabs has agreed to wind T3 with the same wire I am sending out with the mod kits and have used in several past HL2s. They have agreed to keep the lead length to 3mm or less. I will consider it defective and Makerfabs' responsibility to repair if this is not the case.

** Makerfabs has acknowledged that their bag sealing may damage the SMA connectors. They will be more careful.

** The test program will be updated to check for failure to program the FPGA's eeprom. The test program will identify specific parts by schematic/BOM reference number to check if a test step fails. This is to aid Makerfabs technicians. Additional test instructions will be provided for X3, the oscillator on the HL2, which was bad on at least 3 boards.

** The BOM and schematic will be updated to include the modification so that the FPGA can provide the power supply switching clock.

** C84 will be increased to 390 or 420 pF for better high frequency power output as Claudio found. R55 will remain at 120 Ohms.

** The open source logo will be added back to the silkscreen. It was accidentally removed.

** Ground to trace clearance on the N2ADR board will be set again to 0.2mm. The larger clearance may have changed the impedance slightly as noticed by Claudio.

** Thinner PCBs are being considered. Makerfabs says the next standard thinner size from their factory is 1.2mm. The dielectric and distance between top/bottom and inner layers remains the same, so trace impedance should be the same, but there are always surprises. It is the separation between the two inner ground layers that changes. Makerfabs says the specified PCB thickness variation is +/- 0.16mm. The HL2 and N2ADR boards use standard SMA connectors soldered in an edge-launch fashion. The distance from the center pin to two side pins is 1.425mm on these connectors. It is a pressure fit on current PCBs, but may be too loose on a 1.2mm thick PCB. The enclosures that specify 1.6 or 1.7 slot thickness work fine for even thick 1.6mm PCBs. Maybe we should keep the 1.6mm board thickness and those worried about an enclosure which might have too tight of a slot should just pay a few dollars extra for the 1.7mm speced enclosures. There are links on the wiki page.

** Makerfabs could install a small heatsink on the AD9866. This would be the smaller RPi size linked to on the wiki page.

** Makerfabs could include a power connector for those who don't have a 12V power supply with barrel connector.

** Makerfabs could include front and rear panel PCBs. Thanks to Roger for working on this! We've talked about this for several years and there is an open issue on github.

** CL1 and CL2 on the HL2 are currently not installed. I am considering providing a cutout on the PCB for these connectors similar to what is on the N2ADR board so that only a round hole is need on the front panel. 

** Combining the N2ADR and HL2 is an option we considered, but from the build8 experience there would be more failing boards if they were combined. It is helpful to mix and match the two boards when repairing. Build9 will keep the boards separated.

73,

Steve

kf7o

[in3otd]

Hello,

** C84 will be increased to 390 or 420 pF for better high frequency power output as Claudio found. R55 will remain at 120 Ohms.

If the transformer is wound correctly, with the right wire and short leads, this may not be necessary. I've done some measurements on my H-Lv2b8, with a T3 rewound with enameled wire (not recommended but I did not have the right wire) and the default 270 pF for C84 seems to be the best value:


higher values decrease the output power at 30 MHz.
I likely won't have the time to do further experiment with a T3 wound with the recommended wire in the next weeks to see if that works better with a different C84 value; maybe someone with a rewound T3 could try to see if increasing a little C84 improves the output power at 30 MHz? - a relative measurement will be enough.

 

** Ground to trace clearance on the N2ADR board will be set again to 0.2mm. The larger clearance may have changed the impedance slightly as noticed by Claudio.

 The stackup and PCB material used were likely also different from the prototypes done by OSHPark; I'm not sure if we have detailed info about the stackup and material used for the build8 filter PCBs but a clearance lower than 0.5 mm should help to bring the impedance closer to 50 ohm. Note that in the filter board the reference ground plane is not the second layer but the third one, so the dielectric thickness to the ground plane is likely around 1.4 mm and a small top layer ground clearance will then influence the traces impedance.

73 de Claudio, DK1CG / IN3OTD

[in3otd]

I was also wondering if it would make sense to swap the fwd/rev power analog outputs on the filter board, so that PowerSDR users won't have to swap the connections themselves - and it will still work fine for Quisk and SparkSDR users.
It's true that the fwd/rev lines meaning on the filter board depends on how the sensing toroid is connected but it seems to me that currently the only way to have the connection expected by PowerSDR is to have the 10 turns winding ends crossing under the toroid, something that IMHO looks a little odd and likely won't be the "default" way used.
On the filter board PCB there's not much space to swap the connections but I think this could be done around R12, with the line coming from R15 passing thru its pads.

[John Williams]

This may be a good idea Claudio to get the SWR sensing correct. However, I am observing that the N2ADR circuit is about 10X "hotter" than the Alex circuit. I note this by running the Alex circuit that I used for HL1 to the extra connector on the N2ADR board. Spark shows 2 watts as .188 or thereabouts. I agree with Steve that PowerSDR may not have enough capability to scale 10X. Someone needs to manually rewind and test this first.

For those that are following - I recommend that you wait for the new Dick Faust K9IVB boards to be tested. They will be correct for the original Alex circuit and be a better match for PowerSDR users. For most, it will be more interesting to see the output of a linear vs the input. I have some on order for testing and Dick does also.

John W9JSW

--
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...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/f842a08f-995b-4883-b4a5-a924d6d537a8%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Steve Haynal]

Hi Claudio,

I have a rewound T3 with the extra 150pF. I will try to make some measurements with and without the extra 150pF.

73,

Steve

kf7o

[Harold Fizgerald]

looking forward to the next buy de w5mf

[Christopher KB3CS]

* existing board thickness fits well in readily-available no-special-order case, however a thinner board gives the opportunity to solder copper shims to board edge trace - particularly where the PA transistors are and otherwise near the board corners. board would then still fit 'standard' case slots.

* while ideally a heatsink is black anodized for maximum emissivity and fins/posts tall enough to make full use of available space above the A/D chip, wouldn't having to install it be a cost factor? bought in bulk and shipped as a spare part ("u-install-it") drops the cost?

* supplying a coaxial connector with each HL2 is a good idea to me. my net 'stock level' change would be zero, especially if you were to change to the next thicker center pin to reduce supply voltage drop in XMIT. not that i'm suggesting it. existing coaxial connector is widely used.

* the front+rear panel pcbs are looking good and seem nearly ready (some design adjustments underway already). would it be a step too far to consider optionally shipping a case, too?

* wouldn't combining the boards also complicate adding thermal compound under the PA? the corner hole just barely manages to serve as an access port to the case slot allowing compound to get smeared under the PA as the last couple centimeters of the board goes into the case. as is, it's easier to separate the two boards to get the compound smeared under the right spot.

 - 2m (base 25) -

On Sunday, July 14, 2019 at 1:40:57 AM UTC-4, Steve Haynal wrote:

Hi Group,

It is time to start working on the next group buy, build9. Here is my list. Any feedback is appreciated. Many of these items are tracked on the github issues page.

[...] 

** Thinner PCBs are being considered. Makerfabs says the next standard thinner size from their factory is 1.2mm. The dielectric and distance between top/bottom and inner layers remains the same, so trace impedance should be the same, but there are always surprises. It is the separation between the two inner ground layers that changes. Makerfabs says the specified PCB thickness variation is +/- 0.16mm. The HL2 and N2ADR boards use standard SMA connectors soldered in an edge-launch fashion. The distance from the center pin to two side pins is 1.425mm on these connectors. It is a pressure fit on current PCBs, but may be too loose on a 1.2mm thick PCB. The enclosures that specify 1.6 or 1.7 slot thickness work fine for even thick 1.6mm PCBs. Maybe we should keep the 1.6mm board thickness and those worried about an enclosure which might have too tight of a slot should just pay a few dollars extra for the 1.7mm speced enclosures. There are links on the wiki page.

** Makerfabs could install a small heatsink on the AD9866. This would be the smaller RPi size linked to on the wiki page.

** Makerfabs could include a power connector for those who don't have a 12V power supply with barrel connector.

** Makerfabs could include front and rear panel PCBs. Thanks to Roger for working on this! We've talked about this for several years and there is an open issue on github.

** CL1 and CL2 on the HL2 are currently not installed. I am considering providing a cutout on the PCB for these connectors similar to what is on the N2ADR board so that only a round hole is need on the front panel. 

** Combining the N2ADR and HL2 is an option we considered, but from the build8 experience there would be more failing boards if they were combined. It is helpful to mix and match the two boards when repairing. Build9 will keep the boards separated.

[...] 

[Christopher KB3CS]

does pin jumpers with caps make any sense here?  while we wait for an update to PowerSDR, that is?

six pins, 'zig' or 'zag' ..

[Jean pierre Mallet]

[Christopher KB3CS]

with the design rule of "radio in other room - once configured and connected, all functions by remote" in mind, would it be useful to turn the on-board TRS jack into a TX inhibit 'switch'? i know there's provision on the board at CN8 for that. but once the HL2 is in a case with end caps/panels .. short either or both T, R with S and get TX inhibit? or even get fancy with the two bits available and make one TX inhibit and the other PA inhibit?

 - 34 (base 23) -

[Steve Haynal]

Hi Chris,

There is already a TX inhibit on the board and in the gateware. See CN8 on the schematic.

73,

Steve

kf7o

[Christopher KB3CS]

yes, i did specifically mention CN8 in my previous message.

however, once the HL2 is all boxed up, that's less convenient for a radio which might in theory be in a very remote location with limited visits on-site and minimally trained humans to make configuration changes.

once boxed up, best to remain that way as the use shifts from highly trained nearby operator to an over-the-Internet operator of unproved license status.

but now i come to write this, the thought occurs the pcb front endcap could be modified to expose an 'extension' for CN8. 

still, employing the TRS jack is far more robust than exposing a pair of 0.254mm pins on the front panel, yes? 

 - 45 (base 17) -

[Christopher KB3CS]

2.54mm pins

[Steve Haynal]

Hi Chris,

Sorry for not reading your first post carefully. People should be able to add a TX inhibit to the panel as you describe if they desire. CN8 will inhibit TX and the PA if grounded. I'm not convinced I want to make or recommend this as a standard feature, but most of what is done with the HL2 is very experimental and people are welcome to experiment.

73,

Steve

kf7o

[Steve Haynal]

Hi Claudio,

I have tried the extra 150pF on both a HL2b3 and HL2b8. In both cases I see better power output on 15-10M with the extra capacitance. I am measuring power output after the N2ADR filter board. Maybe this explains the differences in our measurements, but I would like to optimize the HL2 for connection to the N2ADR filter board.

270pF Output

38.1dBm 18.1MHz

37.5dBm 21.07MHz

36.6dBm 24.95MHz

37.2dBm 28.1MHz

270+150pF Output

38.2dBm 18.1MHz

37.7dBm 21.07MHz

37.6dBm 24.95MHz

37.9dBm 28.1MHz

37.3dBm 30MHz

I tried with and without two capacitors in parallel to try and measure the effects of lower ESR, but there was no difference.

I think I will change C84 to 390pF for build9.

73,

Steve

kf7o

[Steve Haynal]

Hi Group,

I have posted Hermes-Lite 2.0 build9 design files on github:

https://github.com/softerhardware/Hermes-Lite2/tree/master/hardware/hl/releases/hl2p0build9

The readme on github contains the list of changes. I'd appreciate any review and feedback. The easiest way to review the PCB is to used a gerber file viewer such as gerbv.

I am working on the N2ADR filter board build9 release.

73,

Steve

kf7o

[Steve Haynal]

Hi Group,

I have posted N2ADR filter board build9 design files on github:

https://github.com/softerhardware/Hermes-Lite2/tree/master/hardware/companions/releases/n2adr_e5b9

The link above takes you to a README that lists all the changes.  I'd appreciate any review and feedback. The easiest way to review the PCB is to used a gerber file viewer such as gerbv.

There was a suggestion to swap FWD and REV power to match PowerSDR. I did not make this swap on the PCB. Instead, I will swap the sent value in the gateware.

73,

Steve

kf7o

[Friedrich Schäuffelhut]

Hi Group,

I am interestedin this group buy.
How does it work?
Should I just order on makerfabs right now?

Is this the correct link to the items https://www.makerfabs.com/index.php?route=product/search&search=hermes+lite+2

Thank you

Friedrich

K2CI/DD2CI

[Vladimir Huzevka]

I would love to order one too.  I have the original Hermes (ANAN-10) and I am very keen to test them on the air side by side.  I want to see how much difference there is between the 16-bit and 12-bit ADCs.  If anyone knows when the next run will be then please do let me know ;o)

73

Vlad

VK2IM

[Steve Haynal]

Hi Vladimir and Fredrich,

Please monitor this list for the next group buy. There will be an announcement here and Makerfabs will list units in stock. The price may be adjusted at that time. Makerfabs will keep the group buy open for several weeks. If you purchase within that period, Makerfabs will build and send you a unit.

73,

Steve

kf7o

[rentwist]

Hi Steve,

Here is another alternative to get the power sensing swapped.  The N2ADR board has two holes in each of the pads for the toroid winding.  What I did is remove the toroid, turned it 18 degrees and reinstalled it using the second set of holes (that were unused previously).  Now the power sensing is corrrect.  The next build could install this way and the problem solved that way.  Just  thought.  By the way, the readings were reversed in PiHPSDR as well.

73,

Robert, WA2T

[Bob A. Booey]

Turned it 180 degrees I meant to say!

--
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...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/62306905-0d4b-4a89-a9c9-59b5e31d03d7%40googlegroups.com.

[Steve Haynal]

Hi Robert and Group,

Although this is a way to make PowerSDR see FWD and REV properly, I don't recommend this solution. There are already 100+ units in the field with the pickup is wired as is. The next build will wire the pickup in the same way. I don't want hundreds of people rotating this pickup. Instead, I will release new gateware this week that swaps the values in the gateware. This is part of the efforts for build9. For SparkSDR and Quisk this will not matter as they dynamically switch FWD and REV based on which is returning larger values. This will only help the old set in its ways PowerSDR. So the correct wiring is as it is currently wired.

The scaling is still not correct for PowerSDR. Jim N2ADR did not design the filter board for use with PowerSDR. For best results, people should switch to Quisk or SparkSDR. There is even calibration for the FWD and REV readings in both of these software so that you ca have the most accurate readings. This is how I use the HL2.

73,

Steve

kf7o

To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite+unsubscribe@googlegroups.com.

[Steve Haynal]

Hi Group,

I have submitted all the build9 design files to Makerfabs. They will include front and back end plates for 55mm tall enclosures, but the enclosure must be ordered separately. I went with labeling readable in the "vertical" unit orientation as Roger introduced.

I expect that Makerfabs and I will agree on a final price this week and then Makerfabs will list units as in stock. As before, we must have at least 25 orders before Makerfabs will commit to a small batch. Makerfabs will keep the group buy open for several weeks after 25 orders are received in order to increase the volume.

I will not be handling any customer repairs or returns for build9. Makerfabs will do everything.

I do not work on the Hermes-Lite 2.0 project to make money. My payment from Makerfabs for this build will be 1 working HL2, 3 working N2ADR boards, and any leftover units that they cannot repair. I am also requesting 30 panel sets which I will send to previous group buy members for basically the cost of shipping and handling.

[Amogh Desai]

Hi Steve,

Could you also request MakerFabs to list the front and back panels on their site with their Airmail shipping?

Amogh

VU3DES 

[Christopher KB3CS]

after a couple rounds of calibrating my FoxDelta SWR meter, the PowerSDR readout looks like 10x actual to me. 

close enough to exactly 10.00 times that i am completely fine with it.

  - 1u (base 43) -

On Monday, August 26, 2019 at 12:56:03 AM UTC-4, Steve Haynal wrote:

Hi Robert and Group,

[...] 

The scaling is still not correct for PowerSDR. Jim N2ADR did not design the filter board for use with PowerSDR. For best results, people should switch to Quisk or SparkSDR.

[...]