Other clock frequency

[Art]

Dear Group and Steve. 

I made Hermes-Lite 2 for myself and my friends. In some I did not use versa clock 5P49V5923 and  used a 76.8 MHz oscillator. But a generator for such a frequency is not often available to buy. Can I use a generator on a different frequency? For example, at 75 MHz or another frequency? What is needed for that?

Thanks for the answer.

73, op. Art

[Alan Hopper]

Hi Art,

There are only a small number of frequencies that are easy to make work in the gateware as it has to ultimately be decimated down to 48K, the gateware probably has the remains of previous slower clocks that you might be able to use but it is probably not a good way to go and you cannot easily make it work with just any clock.

73 Alan M0NNB

[Art]

Thanks Alan. I understood. I know a little about the Quartus software. Therefore, if it is not that difficult, I could fix it in the Quartus software

пятница, 26 марта 2021 г. в 18:27:22 UTC+2, ahop...@googlemail.com:

[Steve Haynal]

Hi Art,

The choice of an oscillator for an SDR needs to consider phase noise as that can affect performance. See

https://github.com/softerhardware/Hermes-Lite/wiki/Clock-Phase-Noise

As Alan said, there are some other frequencies supported by the gateware, but currently you must select an oscillator frequency which is a multiple of 8*384kHz.

Perhaps the easiest alternate oscillator is to still use the 38.4MHz oscillator from the BOM but bypass the Versa5 clock generator. Short CL3 and CL7, remove J13 and U6. A change in the gateware (one number changed) is required so that the AD9866 does frequency doubling. I have one HL2 with a damaged U6 which I use in this way. It works fine. Also, the Radioberry uses the same 38.4 MHz oscillator in this way.

73,

Steve

kf7o

[Art]

I am glad for your answer, dear Steve. Doubling the frequency in the ADC itself seemed to me an interesting solution. Since where I live it is difficult to buy Versa5. Please tell me in which file and what changes need to be made to enable the internal frequency doubler in the ADC? Thanks for the help.

73, op. Art  

воскресенье, 28 марта 2021 г. в 20:29:11 UTC+3, softerh...@gmail.com:

[Steve Haynal]

Hi Art,

Search for the parameter BYPASS_VERSA in the RTL. In particular, the line below from ad9866ctrl.v configures the AD9866 to double the frequency:

initarray[4] = (BYPASS_VERSA == 1) ? {1'b1,8'h36} : {1'b0,8'h00}; // Address 0x04,

73,

Steve

kf7o

[Art]

Hi Steve. 

Thanks for the answer

Do I understand correctly what needs to be done (BYPASS_VERSA == 2), and not change the other data? Please see the picture.

And there will be a second question:

Please tell me what were the Set TX Buffer Latency and PTT Hang Time values ​​in gateways 6.2-6.6?

Now I started using gateways of 7 versions, I saw the clicking of the relay. I am trying to set up via hermeslite.py  (python module) but haven’t gotten good results yet. Everything was perfect with gateway versions 6.2-6.6. Everything worked via LAN / WI-FI with different networks and routers. Also, everything worked fine through VPN. My friends from different countries connected to my network without any problems and used remote control.

понедельник, 29 марта 2021 г. в 02:33:21 UTC+3, softerh...@gmail.com:

[Alan Hopper]

Hi Art,

as Steve has probably gone to bed now, I believe you need to set BYPASS_VERSA = 1; here https://github.com/softerhardware/Hermes-Lite2/blob/e2d53bdb9e40e4d93b61c00aecf8c63ee51abc93/gateware/rtl/hermeslite_core.v#L138

73 Alan M0NNB

[Steve Haynal]

Hi Art,

You should set this parameter to 1 at the top level hermeslite.v of a new variant. I checked in my hl2b5up_bypassversa variant:

https://github.com/softerhardware/Hermes-Lite2/tree/master/gateware/variants/hl2b5up_bypassversa

I don't remember what or even if the buffer latency and hang times were in gateware 6.2-6.6. All revisions are in github. You can check past versions with git and find any differences.

73,

Steve

kf7o

[Art]

Hi Steve, Alan and Group,

thanks for the help, everything worked out.

I have not found a code in version 6 that makes a buffer for TX. I'll try to put 0 ms or try to remove the code.

вторник, 30 марта 2021 г. в 07:30:50 UTC+3, softerh...@gmail.com:

[Art]

Dear Steve and the Group. Yes, TX Buffer Latency and PTT Hang Time is a very nice feature. Of course, I managed to minimize the clicks of the relay. This indicates the normal quality of my network. But since I solder HL2 to my friends, they have a lot of questions about clicks. Even if they clicked once during a long TX, they ask questions.

In my new HL2 assemblies, I want to add the AK4951 codec, and only new gateware works there.

Me and my comrades and friends loved the HL2, and the new gateware will be a good update for their transceivers.

Please, tell me, is it possible to somehow disable the PTT HUNG TIME parameter in the code, so that the relays stop clicking? I know about the required value of TX latency buffer and there is no need for PTT HANG TIME.

May think about adding on / off relay clicks to the module hermeslite.py?

For me personally, turning off the clicks in the gateway is enough.

P.S. I was unable to completely disable clicks by changing the Set TX Buffer Latency and PTT Hang Time parameters. I've tried different options. Yes it works. But I want a complete shutdown.

Thanks for the help.

73, op. Art

вторник, 30 марта 2021 г. в 11:18:33 UTC+3, Art:

[Steve Haynal]

Hi Art,

There has always been PTT hang time and TX buffer latency settings in the gateware. The newer gateware makes these settings more accessible. They are and always have been inherent to the gateware design and can't be turned off. In the past the TX buffer latency has been about half of the buffer, so if you set it to ~40ms it should behave similar to past gateware. The intent of recent gateware is to be as good as or better than older gateware. If you or your friends are experiencing clicks, I'd be interested in seeing more debug data. Please try this gateware:

https://github.com/softerhardware/Hermes-Lite2/tree/master/gateware/bitfiles/archive/20201121_72p6

And use hermelite.py to collect data as described in this thread:

https://groups.google.com/g/hermes-lite/c/91uSCsAmQp4/m/Q2vgfnLpAwAJ

Please post your .vcd files and we can try to debug any issues.

73,

Steve

kf7o

[Art]

Hi Steve.

I can not start the debug module. Everything can be seen in the pictures.

To start the hermslite.py module, I manually entered the ip address of the transceiver in the Jupiter notebook.

Python - the latest version for Windows 10 x 64

среда, 31 марта 2021 г. в 06:59:46 UTC+3, softerh...@gmail.com:

[Art]

And I understand that TX delay should be in the gateway. I would just like to hide the clicks of the relay. I have a bad wi-fi, I test on it. So on gateway 6.6 - there are no clicks, but on 7.2 - it clicks constantly. The fact that the transmission signal will be distorted is understandable in both cases.

среда, 31 марта 2021 г. в 08:47:49 UTC+3, Art:

[Steve Haynal]

Hi Art,

Sometimes hermeslite.py won't discover a HL2 if there are mutliple network interfaces. Try substituting your known HL2 IP address instead of 255.255.255.255 in hermeslite.py around line 88:

sock.sendto(msg, ('255.255.255.255', 1025))

I think you don't like the newer behavior when the TX buffer runs out of samples. In older gateware, the HL2 would stay in transmit even if the TX buffer was emptied until told by the software to go to receive. In newer gateware, the HL2 begins the process of going to receive (counting PTT hang time) when the buffer empties. In both cases your transmit signal is corrupted because the buffer is empty and the HL2 transmits quiet. In the old case, the gateware was more complicated and could enter a state of prolonged and unintended transmit. In the new case, the gateware is simplified, there is no prolonged and unintended transmit, plus the user is made aware of buffer empty problems by seeing and hearing relays going back to receive. I don't think relay clicking is your problem, but rather not being able to keep the TX buffer consistently full of samples. If you want to allow the longest latencies set your PTT hang to 31ms and your TX buffer to 50ms or 60ms, just over half the current TX buffer length. Any remaining relay clicks you hear just mean you have a bad TX signal with gaps of missing signal and that is the root problem which should be fixed.

[Art]

Hi Steve.

Thanks for the help . You understood me correctly. The clicks do not bother me. But people who use my designs will not be able to understand even if the relay clicks once a month. And they will have many questions. Many of them are unaware of the technical details and are ordinary radio operators. I will try your solution. 

четверг, 1 апреля 2021 г. в 07:15:25 UTC+3, softerh...@gmail.com:

["Christoph v. Wüllen"]

As far as I can see it just should not happen that the TX buffer is empty
during transmit,
unless there is some fault in the SDR software or in the network
connection.

So we are talking here how to recover from an error condition.



> Am 01.04.2021 um 13:14 schrieb Art <cqdx...@gmail.com>:
>
> And Steve. I am interested in your opinion.
> For example, the size of the PTT hang = 31 ms.
> Such a situation: the buffer is empty, the relay is de-energized for 32 ms.
> But the inertia of the relay coil can be quite large, and the processes occurring in the mechanical part of the relay can distort the transmission signal. What would not have happened if the PTT off signal was transmitted through the openhpsdr protocol from the program.
>
> четверг, 1 апреля 2021 г. в 09:05:44 UTC+3, Art:

> --
> 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/dc4fba4a-4428-40ee-b58c-36aca2a358b1n%40googlegroups.com.

[Art]

Yes, I understand. But my hobby colleagues and I use VPN and work from different places. Collisions are possible there. The unpredictability of jitter over time can still play a decisive role.

четверг, 1 апреля 2021 г. в 15:18:03 UTC+3, "Christoph v. Wüllen":