6M Support

[John Williams]

Alan,

Is this undersampling support also valid for TX? Should we now be
considering support for 6M? The PA will support 6M if we redesign the
roofing filter signal path on the PA and the RX filter on the 1.2 and
1.3 frontends.

What happens to the TX reconstruction filter?

John

[Alan Hopper]

John,

whilst I never really considered tx, I think the code will currently set the correct transmit frequency but will probably swap usb and lsb, this is easily fixed if there is a need.  I've not given this much thought but I guess the reconstruction filter needs to be a bandpass for ( clk/2 + a bit) to ( clk - a bit), there are probably all manner of other issues that others will have a better view on.

73 Alan M6NNB

[Alan Hopper]

John,

not sure what the interpolation on the dac output will do, it might well remove the signal we want, all good brain food.

73 Alan M6NNB

[Takashi K]

Hi all,

I'm interested in undersampling Tx function. According to "Spectrum Analysis Intro" on Hermes-Lite Wiki, the strongest harmonics are at (2Fs - Ftx) and (2Fs + Ftx) when DAC interpolation is enabled. If target frequecy is 50.2MHz, Ftx is 97.256MHz, not realized frequency. So, Interpolation should be disabled ?

73, Takashi JI1UDD

[Takashi K]

Hello all,

I tested the transmission using undersampling today.

Please see the attached file. A better filter is needed.

Alan, I also tested transmitting SSB mode by using your software.

Tx frequency was matched to the setting on the panel. but I checked only 50.200MHz.

I set USB mode on the panel, but I heard LSB sound from my reciever (IC-703). so it's necessary to swap USB and LSB on Tx.

73, Takashi JI1UDD  

[Alan Hopper]

Takashi,

great work. I've just put up a ver 0.131 at http://www.ihopper.org/radio/ which should fix the lsb/usb stuff.

73 Alan M6NNB

[Takashi K]

Alan,

Thanks for the new version. The problem was fixed. Everything works well !

Today, I got 20dBm Tx output on 6m with using 2-pole BPF and PZT2222A amplifier.

But I cannot check whether 23.528MHz signal is suppressed enough or not, because of my equipment limitation.

73, Takashi JI1UDD

[Steve Haynal]

Hi Takashi,

Very interesting experiment. Did you disable interpolation in the firmware? It is now quite easy to disable interpolation, but full duplex will not work if interpolation is disabled. People who pursue 6M TX via undersampling will definitely need alternate bandpass filtering.

73,

Steve

KF7O

[Takashi K]

Hello Steve,
Yes, I needed to disable interporlation. I use the modified 20151018 firmware. In the latest firmware, use initarray_nointerpolation that is replaced the value of Address(0x7) with 8'h20 for Rx LPF off, I think.
I'm not aware of  the limitation, thanks. Now my interest about undersampling Tx is whether IMD can be improved or not by using TxDAC. Does anybody have any good ideas (circuits) ?
73, Takashi JI1UDD

[Takashi K]

Hi all,

I checked the IMD of 6m Tx signal with using TxDAC.

It's under -60dBc but It's difficult to keep the quality through amplifier.

73, Takashi JI1UDD

[Steve Haynal]

Hi Takashi,

Too bad that full duplex doesn't work with 6M TX as then we could use pure signal.

73,

Steve

KF7O

[Takashi K]

Hi Steve,

For using pure signal on 6m, Tx uses frequcncy converter and Rx uses undersampling. How about this idea?

73, Takashi JI1UDD

[Steve Haynal]

HI Takashi,

Sounds interesting. I'm not sure 6M support is that important to most, but it is always interesting to see new ideas.

73,

Steve

KF7O

[Takashi K]

Hi Steve,

Sorry for my late reply.

I think that the Tx frequency converter can be added to H-L v2 system easier than undersampling Tx. 

But I prefer undersampling Tx because H-L is DDC/DUC system. Now I'm considering to make 6m frontend for H-L v1.22.

73, Takashi JI1UDD

[Alan Hopper]

Takashi,

I like the idea of a 1.2 6m frontend, it would make especially good use of peoples non duplex v1.2 & 1,21s once they all have v2s,  I'm intrigued how high in frequency we can go on undersampled transmit.

73 Alan M6NNB

[Steve Haynal]

Hi Takashi,

I concur. A 6M frontend for the v1.2 series would be interesting. All the kicad files to get started are on github.

73,

Steve

KF7O

[Takashi K]

Hello Steve, Alan and all,

This is my recent report of 6m frontend.
I designed 6m frontend using KiCad, and ordered the board to FusionPCB two weeks ago.
Yesterday, I received it. I assembled immediately.

Now, the first assembled board works well.
Tx power is about 1dBm, less (-4dB)  than my previous experiment.
Spurious except harmonics of 50MHz(100MHz, 150MHz,...) seems to be less than 60dBc, including the original 23.528MHz.

But I made a mistake, source and drain of Q2 and Q4 were reverse, when I drew the schematic.
Q2 and Q4 are power switch for Tx and Rx circuit. T12V for Tx buffer amp can be supplied from external power circuit.
So, It's not serious.
( My counter measure is, bend the lead of FET reversely and install with rotation. refer to photo)

I have seven extra PCBs. I can distribute them if you have any interest.
I attached the schematic v1.1, the bug is fixed.

73, Takashi JI1UDD

[Steve Haynal]

Hi Takashi,

That looks very nice! I am glad to see several people build different frontend cards. This was original purpose of the pcie connector -- allow easy experimentation with small frontends. 

Have you had any QSOs on 6M with this setup? I might build one later but now I want to get HL2 done. There may be others on this list with interest in 6M.

73,

Steve

KF7O.

[Alan Hopper]

Takashi,

that looks great. I'd be interested in a board.  If anyone else in the UK or Europe is interested, I'll happily re-post them if that saves on postage. 

Is the 1dBm a typo?

73 Alan 2E0NNB

[Takashi K]

Hi Steve,

I have not yet. Because I'm still looking for good PA for 6m.

I want to achieve output power from 1W to 5W and good IMD.

BTW,  I know you are busy now.

I keep one board for you. Please feel free to request if you need.

73, Takashi JI1UDD

[Takashi K]

Hi Alan,

Please e-mail your address to me. I will send you one. 

I attached BOM.

Is the 1dBm a typo?

Not typo.

But, in my previous experiment, It's 5 dBm.

When I assembled, I adjusted Diplexer and BPF individually.

Also I checked the characteristic of them individually.

The loss of Diplexer is around 0.1 - 0.2dB, BPF's is around 3dB. Each loss is reasonable value.

I guess the characteristic was shifted.

Bucause Diplexer and BPF are connected directry. So It may have affected the characteristic of each other.

I will try to check the characteristic of the combination befor re-adjusting trimmer caps.

73, Takashi JI1UDD

[John Williams]

Takashi-san,

I happen to have an extra board for one of Claudio's other amps. I am willing to send it to you at my cost if you want to experiment with it. I was going to build it but decided to go in another direction. One other ham on this list built one but I do not recall his name at this time. He made some interesting heat sinks out of 10 or 12ga copper wire.

Here is the link to Claudio's page - http://www.qsl.net/in3otd/ham_radio/PD85004_PP_PA_II/PD85004_PP_PA_II.html

Let me know. It is worth it for me to see how you progress on this undertaking. I have a second Hermes Lite 1.22 that could be re-purposed for a 6M rig.

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.
For more options, visit https://groups.google.com/d/optout.

[in3otd]

Hello Takashi,
FYI, I have recently measured the driver you use in your nice 6 m frontend using a better (but still low-cost) transistor, see http://www.qsl.net/in3otd/ham_radio/driver13/driver13.html#2sc5551a ; it should work a bit better better at 50 MHz than the one using the PZT2222A (it does not need the 680 nH inductor to flatten its gain). By the way, the driver could probably be further improved by tuning it to 50 MHz, the original design was intended to be wideband and cover the HF to 30 MHz only.

73 de Claudio, IN3OTD / DK1CG

[Takashi K]

Hi John-san,

Thank you for your advice.

I often watch Claudio's home page that has very useful information for me.

Therfore I know his PD85004 PP amplifier. To tell the truth, I ordered the board to FusionPCB last week. I'm wating for the package now. So you don't need to send me your PCB. thanks.

I also have several RD01MUS1. I will check PD85004 PP and RD01MUS1 PP.

I have already checked 2N2222 Norton amp that connect to 6m frontend (Now 1dBm output). It has good IMD, but gain is around 9dB, not enough.

I think that  driver gain is 20dB and final's is 15dB is suitable.

How about OPA2677 for 6m ?

BTW, Shall I send 6m frontend board to you ?

73, Takashi, JI1UDD

[Takashi K]

Hi Claudio,

Thanks for your advice.

I replaced 680nH with 0 ohm jumper and measured 6m frontend output.
The result is, the output decreased about 1dB, so 0dBm now.

I guess amount of feedback became original design value, so gain decreased.
Therefore the output power seems to be changed a little even if I use 2SC5551A. 
You mean, re-design feedback parameters?
How do you design resistive feed back amplifier always?  about Bipolar Tr case and MOS FET case.

Or please let me know where the reference is.

73, Takashi JI1UDD

[John Williams]

Takashi-san,

I would very much like one of your boards.

OPA2677 is used in the Hermes design and that design supports 6M so this should be viable. Perhaps you can take one of the 1.42 boards and modify/remove the filters and give it a try? I will send you a board.

John

[Takashi K]

Hi John-san,

Thank you for your information.

Yes, I will try to modify 1.42 board and test for 6m driver.

I will e-mail my address later. So please reply your address.

73, Takashi JI1UDD

[Takashi K]

Hi all,

I checked Rx and Tx over all characteristics. There is no problem, I think.

But I got one mystery. see attached document, page3.

73, Takashi JI1UDD

[Takashi K]

Hi Steve,

I have a request.

Could you release new firmware for supporting 6m frontend (undersampling) ?

My idea is simple.

Check spare DIP SW when power up, then set interpolation/RxLPF to 1x/OFF.

The setting of the firmware now I use is fixed. so it's not convenient.

73, Takashi, JI1UDD

[Steve Haynal]

I'll try and do that this weekend, but please send me a copy of your working hermes-lite-core.v.

73,

Steve

KF7O

[Takashi K]

Hi Steve,

Thank you for your understanding.

When I check undersampling operation, I use the firmware that was downloaded from Claudio's git ( late last year, wrong gain mapping story ) and I modified. It's based on 20151018 version.

So, I attached "ad9866.v" and "The point of modification.xls". The excel file is helpful to understand.

73, Takashi, JI1UDD

[Takashi K]

Hello John and Alan,

The boards are on the way. I think it takes 10 days from 7 days. 

I attached the document of v1.0 PCB Errata and filter adjustment.

I re-adjusted trimmer caps today. The output became 6dBm. But the filter characteristics has not been checked yet. 

Diplexer Q is low ,10. Since the resonance is broad, it's difficult to adjust to the best point in combination with BPF.  

73, Takashi JI1UDD

[Takashi K]

Hi all,

FYR.
1. The two peak phenomenon seems to be improved by adjusting the trimmer Capacitor.
2. The expected output power of 6m frontend.

73, Takashi JI1UDD

[Steve Haynal]

Hi Takashi,

I have not forgot about this, but have run into problems when recompiling the RTL. I fixed an issue with gain settings in full duplex, but the RTL has not been stable and enters a high noise floor state after 8-12 hours of operation. I have also updated to Quartus 16.0 and am using a board with an old connector, both of which may be the reason behind my problems. I have committed my changes to github and would appreciate it if you could test the 6M option. To customize the RTL, one should follow these steps:

1. Select the desired revision (full duplex, half duplex, CV, CVC8, CVA9, SDK) in Quartus to match your BeMicro.

2. Make sure the clock frequency and number of receiver (NR) parameters in Hermes_Lite_*.v are set properly for your target.

3. In hermes_lite_core.v around line 280, you must set the initarray assignment to use the new 6M initarray. There are instructions in the RTL comments. This is also the place where you set disable_IAMP if you are building for a board such as the V1.32-V1.42 frontends that use the TxDAC.

The other instructions on the wiki for recompiling the RTL remain the same. You will have to update to Quartus 16.0 if you want to use the latest RTL from github. If you want to use an older Quartus, only update hermes_lite_core.v from github.

73,

Steve

KF7O

[Takashi K]

Hi Steve,

Thank you for taking your precious time.

I haven't known that SDK and CV have fewer DIP SW and CVA9's 3bit is already used.

BTW, Can I confirm current DIP SW assignment ? Because comments in hermes_lite_core.v seems wrong.

  DIPSW[0] : Hermes or Hermes-Lite

  DIPSW[1] : alternative MAC address

  DIPSW[2] : Test Mode ( for CVA9)

If my understanding is correct, I will modify like this ;

    localparam bit [0:19][8:0] initarray = dipsw[1] ? initarray_6m : initarray_regular;

and then I will compile and test 6M option.

73, Takashi JI1UDD

[Takashi K]

Hi Steve and all,

I replaced with new hermes_lite_core.v and tried to compile, but the error occured.

The cause is that dipsw[1] is not constant.

Does anyone know any good workaround ?

73, Takashi JI1UDD

[Steve Haynal]

Hi Takashi,

It looks like we can't dynamically change a parameter. I will have to pass this down as wires later once I return home. You can skip the selection based on the dip switch just for now if you want to test the 6M settings.

73,

Steve

KF7O

[Takashi K]

Hi Steve,

I tested 6M setting with Alan's software and PowerSDR(TUNE mode). Both Rx and Tx works well, but Tx TUNE output decreased from 5dBm to 2dBm.

I wrote the previous firmware into EEPROM again and confirmed this phenomenon.

Is this a reasonable? 

My previous firmware is Claudio's, and I modified ad9866 initial setting for undersampling.

73, Takashi JI1UDD

[Steve Haynal]

Hi Takashi,

Are you using the TxDAC our IAMP output? According to the excel file you sent, I think line 246 in hermes_lite_core.v should be changed to:

    {1'b1,8'h01}, // Address 0x0e, Enable/Disable IAMP 

This will disable the IAMP and may explain the differences in power output. Sorry, but there are just too many configuration variations for me to keep track of, and I'm looking forward to V2 which should have fewer.

73,

Steve

KF7O

[Takashi K]

Hi Steve,

I don't use IAMP output. TXP and TXN from H-L v1.22 are opened on 6m Frontend board.
So I will try your suggestion.  Also I will check the difference ad9866_drive_level setting between your new FW and Claudio's.

Thanks.

73, Takashi JI1UDD

[Steve Haynal]

Hi Takashi,

Okay. Let me know what you find out. 

I just committed code to github that will allow you to select between two initarrays with the most significant bit of the dip switch. This should also be useful to experiment with other initarray configurations. You must reset or power cycle the Hermes-Lite for the change to take effect. Just changing the dip switch is not enough.

73,

Steve

KF7O

[Takashi K]

Hi Steve,

The initarry select function is big advantage for me. Thanks.

Unfortunately, Quartus 15 on my PC was expired yesterday. I will update to Quartus 16 at first. So I need more time.

73, Takashi JI1UDD

[Takashi K]

Hi Steve,

I tested your latest FW today. The new function works well.  Thanks.

But as I mentioned before, dipsw[2] (= DIP SW #3) is used for CVA9's test mode setting .
So I use dipsw[1] in hermes_lite_core.v instead.

ad9866 #(.initarray0(initarray0), .initarray1(initarray1)) ad9866_inst(.reset(~ad9866_rst_n),.clk(ad9866spiclk),.initarray_sel(dipsw[1]),.sclk(ad9866_sclk),.sdio(ad9866_sdio),.sdo(ad9866_sdo),.sen_n(ad9866_sen_n),.dataout(),.extrqst(ad9866rqst),.gain(dd));

This means,

  DIP SW #2 ( = dipsw[1] ) : ON = Normal,  OFF = Undersampling(6M)

So, I think  .initarray_sel(~dipsw[1])  is more suitable, but not checked yet.

Also the diffrent output power issue seems to be fixed. I will check again this weekend.

73, Takashi JI1UDD

[Steve Haynal]

Hi Takashi,

Good point. I will switch to dipsw[1] next time I update the RTL.

73,

Steve

KF7O

[Takashi K]

Hi Steve,

The diffrent output power issue was solved.

There were two causes. One was IAMP enable/disable setting, the other was TxPGA setting.

When using TxDAC output, the latest your FW is perfect by replaced with 

 {1'b1,8'h01}, // Address 0x0e, Enable/Disable IAMP

as you suggested.

So, my frontend 6M output became the following,

      PSDR             TxPGA         TUNE mode   Two-Tone

     DriveLevel       setting          output pwr       IMD3

Min.     0                      0             1 - 2dBm       57dBc

Mid.   18 - 23               7             5dBm            54dBc

Max.  83 - 100           15             9 - 10dBm     48dBc

I checked TxPGA setting value( 0 - 15) by using LEDs on Bemicro CVA9. 

73, Takashi JI1UDD

[Takashi K]

Hi all,

I added 6m PA and AudioCodec board.

The output power is around 3W in CW mode. But in SSB mode, I have to use within 1W to keep IMD3 under -30dBc.

If anyone have any good idea for IMD3 improvement, please let me know.

Rx PreAmp is still in debugging.

73 Takashi, JI1UDD

[John Williams]

Takashi-san,

Perhaps I am confused - is R12 and R13 truly DNI? On my single stage amplifier using the RD16HHF1 instead of the RD06HHF1, I use a 270 ohm R in that position. Also, I use a 100nH inductor instead of the 2.2 ohm gate resistor. We just measured the amp and it is stable well past the 50MHz band.

John - W9JSW

[Joe]

Takashi,

 Details of AudioCodect please. 

 Thanks  Joe

[Takashi K]

Hello John-san,

Yes, now R12 and R13 are DNI, no NFB. Because I want to get the output power over 1 watt at least.

Therefore I adjusted the input impedance to 50 ohm by LC circuit (HPF type).

Before that, I have tried several type(primary and secondary turn) of input transformer. but I could not get the good result.

I often see a pallarel capacitance across output transformer. I have not tried it yet because I don't have a suitable variable capacitance.

Have you ever used this technique ?

73 Takashi, JI1UDD

[John Williams]

Takashi-san,

How much gain are you achieving? Attached is my design, but I have not yet tested it. However it is derived from the PA-2008 design that is also attached. They both use 150pf mica cap on the output transformer.

John

[Takashi K]

Hi Joe,

I refered to ik1xpv homepage and git.

Here is my schematic that is basically same as ik1xpv's.

I connected this board to my CVA9 and modified the latest firmware.

I couldn't use PowerSDR in SSB mode before because VAC cannot work well on my Windows8.1 PC. But now I can use PowerSDR in SSB mode using audio codec instead of VAC.

Do you want to try?

I have extra PCBs.

73, Takashi JI1UDD

[Takashi K]

John-san,

Thank you for your information.

The  gain is around 25-26dB including LPF.

The output of frontend 6m is 9-10dBm, and the PA's is 35dBm.

How about your RD16HHF1 P-P amplifier ?

73 Takashi, JI1UDD

[John Williams]

Takashi-san,

We are seeing around 20db gain on the single mosfet design. I expect a few more DB on the PP one but have not tested it yet. The reason we were considering it was to have less harmonics.

On your amp, why not add a norton preamp stage to gain 8-10db? We did that on the 1.0 version of my PA when we were seeing 10dbm output on earlier versions of the frontend. I think Claudio's page shows some data and would indicate if it can run at 54MHz.

Our current plan uses opa2677 and get a solid 20dbm, thus the 1.1 PA no longer has the norton preamp. Perhaps you should consider that as well. It can provide differential output and simplify the input circutry of your PA. The design is from hermes, that supported 6M band.

John

[Takashi K]

Hi John-san,

> We are seeing around 20db gain on the single mosfet design.
> I expect a few more DB on the PP one but have not tested it yet.
> The reason we were considering it was to have less harmonics.

I see. H-L requires the good broadband amp that has less harmonics and flat gain.

If you get the theory of parallel capacitor, please let me know.

> On your amp, why not add a norton preamp stage to gain 8-10db?
> We did that on the 1.0 version of my PA when we were seeing 10dbm output on earlier versions of the frontend.
> I think Claudio's page shows some data and would indicate if it can run at 54MHz.

I have tested PZT2222A norton amp, OPA2677 and PD85004 PP amp.
They didn't get 1 watt output power in my test condition (frontend 6m drives them), needs final amp or driver amp.
I considered that which is better IMD totally, two stages amp? or one stage amp?
And I chose one stage amp. That's the reason.

> It can provide differential output and simplify the input circutry of your PA.

I think so, too. but it seems difficult for me to realize it as excellent HF-6m broadband amp.
I need to study how to improve IMD around output transformer at first.

BTW, for input transformer, I got the better result with center tap on my PA.
When using input transformer without center tap,

the input amplitude of two FETs became different level, ex. 1.0Vrms and 0.6Vrms.
Just my experience. please check it if you have a chance.

73 Takashi, JI1UDD

PS. I got HELA-10D amplifier for my test environment.
    It's extremely good characteristics. IMD doesn't get worse at all in my test.
    But It is expensive and has high power consumption(= hot).
    See SV1AFN homepage.

[Takashi K]

Hi all,

I received the request about firmware adapted to audio codec by e-mail.

So, I post my modified firmware.

If apply this firmware,

Step1. Check my modified RTL by comparing with the latest official RTL.

Step2. Modify top module (in my case, Hermes_Lite_CVA9.v) and core module (hermes_lite_core.v).

Step3. Define FPGA pin assingment

Step4. Add three files, I2S_rcv.v, I2S_xmit.v and TLV320_SPI.v on Quartus. These files can be download from ik1xpv git.

Step5. Execute Analysis & Synthsis

Step6. Assign FPGA pin by using PinPlanner.

Step7-End.  Fitter - convert to jic

73 Takashi, JI1UDD

[in3otd]

Hello Takashi,
you can find all the theory about the compensation of the transformers response using some parallel capacitance in the old Philips Application Notes
http://www.radio-kits.co.uk/radio-related/Linear_PA/ECO6907.pdf
and in particular
http://www.radio-kits.co.uk/radio-related/Linear_PA/ECO7213.pdf

but I'll say that in practice you just determine the best value experimentally, hi.

73 de Claudio, IN3OTD / DK1CG

On Sunday, June 26, 2016 at 11:04:32 PM UTC+2, Takashi K wrote:
[snip]

[in3otd]

Hello Takashi,
in theory you should be able to get about 4 W CW from that amp and a figure of 4 W PEP (which are really 2 W read on a power meter, 1 W per tone) in SSB sounds reasonable. The 1 W you get in SSB is "per tone" or the total output power?

Feedback could improve a little the IMD3, but you'll have less gain... Is the PA intended to cover also the HF or only the  6 m band?

73 de Claudio, IN3OTD / DK1CG

[Takashi K]

Hello Claudio,

Thank you so much for your information.

I have not read these materials well yet. but they seem to be very useful for understanding the theory.

Actually, as you said, cut and try is needed, I guess.

73 Takashi, JI1UDD

[Takashi K]

Hello again,

I measured the output power using 10dB ATT, 50 ohm load (Tektronix 2W) load and scope.

When I set drive level of PSDR to 100 and use TUNE mode, the power was 3 watts.

But IMD3 of two-tone test (700Hz+1700Hz) was worse than -25dBc.

After that, I needed to reduce drive level under 31 in order to keep IMD3 under -30dBc. In this case the output was nearly 1 watt.

> Is the PA intended to cover also the HF or only the  6 m band?

The maximun output of frontend v1.42 is around 20dBm, but my frontend 6m is around 10dBm. There is 10dB gap.

If one PA covers HF to 6m and the output is 5watt, I will be happy. but this is next step, I think.  

Now I need to swap frontend board on HF and 6m.

73 Takashi, JI1UDD

[John Williams]

It would be good to try the 1.42 board at 6M with appropriate Tx differential filter for that band. Perhaps Jim Ahlstrom can refresh our memory on what spurs occur between 10M and 6M that are problematic? If we allow a spur in that space on a 1.42 frontend that has a 6M filter, perhaps the PA TX LPF filter can remove it for 10M and lower freq's. I am not proposing that we change the 160M to 10M design point per se, but wondering if there are alternate build options for those that desire to operate on 6m?

[Takashi K]

John-san and all,

Unfortunately I don't have enough knowledge to design Tx differential BPF for 6m.

As I mentioned before (7th May), 6m signal from TxDAC is 6.6dB less then fundamental signal theoretically.

But if OPA2677 gain is increased and  the diffrent output level is compensated by TxDAC gain control, the remained issue is switchable diffrential filter, isn't it?

73 Takashi, JI1UDD

[John Williams]

I believe so, or one can have two build options on one board. Build for 160M to 10M or build for 6M.

[Steve Haynal]

Hi Takashi,

You can use Elsie to design a balanced filter. I think people often start with Elsie, but optimize and tune what Elsie produces.

In an earlier post, you mention "Rx PreAmp is still in debugging." Is the AD9866's builtin -12 to +48 db low noise amplifier in front of the RX ADC inadequate for your purposes?

Thanks for all your interesting work with 6M!

73,

Steve

KF7O

[Takashi K]

Hi John-san,

I see. two build option is easier than one board covers 160m to 6m.
But, In both cases the important point is whether we can design good Tx differential BPF for 6m or not.
As 6m signal is image signal and less than fundamental signal, I think we need high Q circuit and to care VSWR at TxDAC output.

73 Takashi, JI1UDD

[Takashi K]

Hi Steve,

Thank you for your information. I will try to use Elsie later.

For Rx PreAmp, when I used my frontend 6m with PowerSDR to hear some QSOs, I felt the sensitivity was lower than my ICOM IC-703.

I have not checked the detail. e.g. check sensitivity vs frequency (HF - 6m) by using good SG such as Agilent (I don't have one).

So I implemented PreAmp before BPF experimentally.

I always use PowerSDR with the following setting (I think this is recommended setting),
  Dither: ON
  Random: ON
  Attenuator: Enable, value: 0
  S-ATT : 0
  AGC   : Med
  AGC Gain: 90-100

The gain of PreAmp is around 11dB. NF is 1.5 according to my reference book (in Japanese). 

73 Takashi, JI1UDD

[Takashi K]

John-san,

Could you teach me about your Power2 and FR-Sense board ?

I'm interested in them for my next stage.

1) For FWD/REV circuit

    Are you using the combination of these boards ?

    Because I saw another FWD/REV pickup coil on your schematic, looks like HARDROCK-50.

    I prefer one BN43-202 pickup coil to two FT43-37.

2) For VFWD circuit

    Is it necessary to implement this circuit on ADC board ?

    Because I could not find ALC circuit in current H-L firmware. 

    ALC circuit is described on Penelope Verilog block diagram.

73 Takashi, JI1UDD

[James Ahlstrom]

Hello John,

On Monday, June 27, 2016 at 7:21:22 PM UTC-4, John Williams wrote:

It would be good to try the 1.42 board at 6M with appropriate Tx differential filter for that band. Perhaps Jim Ahlstrom can refresh our memory on what spurs occur between 10M and 6M that are problematic?

There is a strong spur at 45 MHz when transmitting on 10 meters.  It is at Fclk - Ftx.   This spur is very troublesome, and forced us to use a 7-pole filter as the reconstruction filter.  The expected spurs around 147 MHz would have been much less problem.

Jim

N2ADR

[John Williams]

Takashi-san,

The Power 2 board implements 2 hermes features. VFWD which is sensing of the opa2674 output, max of 500mW. The FWD/REV lines are for remote sensing. In my implementation, I sense the output of my 50W hardrock type amp (actually a ARRLHBC Run 2 amp). The circuit was derived from a K3 sensing design. I chose it because the inductors are trivial to wind. You can see both of these (in earlier configurations) on my qrz.com page.  It also derives 5V and 3.3V for my HL 1.22 board.

To use the boards, if you do  not enable Alex support in PowerSDR, it senses VFWD. If you enable Alex, you see Fwd/Rev function, depending on the setting of the meter pulldown. On the PA tab, you can expose some more data fields and actually see forward reverse and vfwd all on one screen. Very handy.

They also have opto-isolated signals for CW in, CW out and PTT in.

John

[Steve Haynal]

Hi Takashi,

Your gain settings look like +19 dB, which is typical for HF. There is a table on the software wiki. If you feel like you need another 11 dB, what happens when you change the gain to +30 dB according to the table? I'm not sure how yet another LNA preamp will help. One reason for the attenuation in 6M is the sinc envelope of the ADC. You are approaching the sampling frequency. Another reason is that the AD9866 has a builtin ADC. I can't remember if this is enabled or disabled in your firmware.

73,

Steve

KF7O

[Takashi K]

John-san,

Thank you for your explaining kindly.
I understood that VFWD can be monitored on PowerSDR and currently you use the sensing circuit derived from K3.

> They also have opto-isolated signals for CW in, CW out and PTT in.

Nice !

BTW, Do you have these extra boards ?

I think maybe no,...

73 Takashi, JI1UDD

[Takashi K]

Hi Steve,

Thank you for your advice always.

I checked the sensitivity roughly.

As you said,  It was enough on 6M when AD9866 Rx gain was set to +30dB.

I need to change the setting between HF and 6m every time.

It's not convenient. But I think this can be fixed by firmware or software.

73 Takashi, JI1UDD

[Takashi K]

John-san and Steve-san,

I got a hint to convert my current single-ended filter to balanced-ended one from Elsie.

So, I did convert 6m single-ended filter to balanced-ended.
I have simulated and compared each other by using LT-Spice.

The AC analysis result of balaned-ended is mostly same as single-ended. But I haven't experimented yet.

73 Takashi, JI1UDD

[Takashi K]

Hi all,

I have verified the 6m balanced filter. It works well on frontend 142.

This time, I use T25-6 for all inductors.

The OPA2677 gain setting is default, not changed.

The result is,

  6m output :  13dBm , resonable output.

  IMD3        :    -43dBc.

In my case, before this modification,  

My frontend 142's IMD3 at 28MHz was -54dBc, but at 7MHz was -78dBc, very excellent!

73 Takashi, JI1UDD

[Takashi K]

Hello Steve,

I considered Tx filter build option. please refer to the attached pdf.

How? Is it possible?

I think trimmer caps are needed for adjustment.

Another idea is,

connect the current HF-LPF to diplexer output not 6M side, and select either output of LPF or BPF by RF SW, and  switch OP-AMP gain. But in this case, The circuit becomes complex and need more PCB area, and need more time for experiment. Therefore, build option is better, I think.

73 Takashi, JI1UDD

[Steve Haynal]

Hi Takashi,

Thanks for the interesting circuit. I think a build option for 6M is better. I am worried about people having to adjust the trimmer capacitors without much test equipment. I would like a basic build with SMT inductors and fixec capacitors that people can just about pretty much get right every time. A more advanced builder can substitute their hand-wound T25-6 inductor if they like. Currently with HL2, someone who wants to experiment with 6M would have to use their own daughter board that replaces for the filters and op amp in the v1.42 board. Given the work someone must do (use undersampling, limited BW of AD9866, etc.), I am concerned the HL will never be a great performer on 6M. Also, for TX with the IAMP, there is a nasty spur at Fs-Ftx. I don't recall if this was eliminated when switching to the TxDAC. The results are somewhere buried on this list. How do spurious emissions look when transmitting on 6M?

73,

Steve

KF7O

[Takashi K]

Hi Steve,

Here is my result. I think there is no problem.

But there is a risk. the performance will be worse by using lower Q compornents.

Also, as you mentioned, we cannot guarantee 6m performance because user have to adjust by themselves.

For 6m setting,  I am satisfied with HL 1.22,  my current 6m frontend and powerfull CVA9.

So, even if  build option is dropped, I do not mind.

73 Takashi, JI1UDD

[Steve Haynal]

Hi Takashi,

Thanks for the measurements. It is good to see that it is clean and does not suffer from the Fs-Ftx spurs. 

73,

Steve

KF7O

[Takashi K]

Hi Steve,

Unfortunately, I found a serious issue. That is in-band spur when Fs = 76.8MHz setting.

See attached file.

I think 6m build option should be dropped.

73 Takashi, JI1UDD

[Steve Haynal]

Hi Takashi,

Thanks for making these measurements. With the VersaClock on the HL2 we will be able to change Fs dynamically. So we could have the firmware change to 73.728 MHz for 6M use. 76.8 MHz is preferred as it is exactly 2x 38.4 MHz, a common frequency to find oscillators for that does not need any non integer PLL multiplication to achieve 76.8.

I do think (and believe you would agree) that a 6M option is a pretty advanced enhancement or build option for the HL2 that only a few experienced builders would pursue. I completely understand if you think a 6M build option should be dropped. Still, I have found your work on 6M very valuable and interesting and do not want to see it forgotten. Do you think we can capture all your work on a single 6M wiki page, or even your personal HTML or QRZ page that I can link to? I can start a wiki page for you that you can edit if you like. In the future, someone will come along and want to try 6M and your experience will be very instructive.

73,

Steve

KF7O

[Takashi K]

Hi Steve,

> With the VersaClock on the HL2 we will be able to change Fs dynamically.

Yes. But it is necessay to use undersampling transmitter with understanding its useage condition well. I think it is for expert but it's fun for amature users that make their own rigs as well.

> Do you think we can capture all your work on a single 6M wiki page, or even your personal HTML or QRZ page that I can link to?

Yes. I have started preparing my git yesterday. But need more time because I'm beginner for git and html.

73 Takashi, JI1UDD

[Takashi K]

Hi Steve and all,

I released the materials of my Hermes-Lite enhancement(6M and Keyer) on my git, 6M branch. 

Here !

73 Takashi, JI1UDD

[Steve Haynal]

Hi Takashi,

It looks very nice. Thanks. I will add your link to my pages this weekend.

73,

Steve

KF7O

[Steve Haynal]

Hi Takashi,

I have linked to your page from main the Hermes-Lite wiki, Google Group and main Github pages.

Thanks and 73!

Steve

KF7O 

[Takashi K]

Thank you so much, Steve.