Radioberry Solid State QSK TR Switch

[rentwist]

Hi Johan and Group,

Here is a Solid State QSK TR Switch that I designed a few years back for another SDR.  It provides about 42 dB of PA input/output isolation and a little over 1dB insertion loss for RX and TX.  The design is adapted from the Hardrock-50 QSK TR Switch which has its roots in the Ten Tec Argosy QSK TR Switch.

Anyway, it occurred to me recently that this switch might be a perfect fit for the Radioberry, plan to test it with a QRP-Labs QSX PA and the Radioberry very soon.

Here is a picture of an assembled board (and I have a bunch of these PCB's!):

What do you think?

73,

Robert, WA2T

[pa3gsb]

Robert

That is  nice idea; do you have a schematic. 

Please let us know the result of your experiment.

Johan

PA3GSB

[Bob A. Booey]

Hi Johan,

Yes, I have all the design info, I will have to dig it out.  The QSK board PCB (and schematic) were done in Eagle, this was done before I learned Kicad.

73,

Robert, WA2T

--
You received this message because you are subscribed to the Google Groups "Radioberry" group.
To unsubscribe from this group and stop receiving emails from it, send an email to radioberry+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/radioberry/fd28b055-5dc1-4932-85db-e0536c69e4d0%40googlegroups.com.

[rentwist]

Hi Group,

Sorry for the delay, finally struggled through digging out the schematic on my previous, slow, bogged down, almost unusable laptop!

Here is the QSK board schematic.

73,

Robert, WA2T

On Friday, May 15, 2020 at 7:33:26 AM UTC-4, rentwist wrote:

Hi Johan,

Yes, I have all the design info, I will have to dig it out.  The QSK board PCB (and schematic) were done in Eagle, this was done before I learned Kicad.

73,

Robert, WA2T

On 5/14/2020 2:31 AM, pa3gsb wrote:

Robert

That is  nice idea; do you have a schematic. 

Please let us know the result of your experiment.

Johan

PA3GSB

--
You received this message because you are subscribed to the Google Groups "Radioberry" group.

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

[rentwist]

P.S.  I will post meaningful port names/connections in a little while.

[rentwist]

Hi again Group,

Looking at the design, it is/was intended for a PA with a single input and output whereas the Radioberry has a separate RX and TX port.  This should simplify the design (less components/switch elements needed).  Will advise on this once I've worked it out.

73,

Robert, WA2T

[rentwist]

Hi Group,

Okay, here is how to use the QSK TR Switch with the Radioberry.  Change the design from a SPDT/SPDT topology to a SPST/SPDT topology.  To do that:

Referring to the schematic, delete D3, C4, L3 and C8, thereby eliminating the "AMP_IN" port.

Connect the PA to the TX output of the Radioberry (directly, no switching) and (again from my schematic) RY1 pin O1 to Radioberry RX, RY1 pin S1 to the PA output and RY1 pin P1 to Antenna.

RY1 does not exist in reality, just the footprint for the relay that was to be replaced by the QSK TR Switch board.

I will test this with my RB and PA and report back.

73,

Robert, WA2T

[rentwist]

P.S. RY1 pin 16 is control and is high active.

[rentwist]

Hi Group,

I have gotten the QSK TR Switch working with my Radioberry and the QRP-Labs QSX Power Amplifier!

The only other changes I needed to make to the schematic besides the previous post about deleting some components (actually, I did not even delete these, I left them in place) was to change Q1 to a 2N7002 as the 3.3V high from the FPGA was not enough to turn-on the 3N42 MOSFET.  Also, I removed R5 and changed R7 to a 10K ohm.  R5 needed to go in order to delete the divider and I decided to change R7 to a 10K ohm to be kind to the unbuffered FPGA PTT output.  Works well!

Upon testing, I noticed that RX recovery from TX is very (very) slow in CW mode, but when using Tune RX is right there when Tune is released.  This may have something to do with the recent Hermes Lite 2 gateware which was used as the basis for the latest Radioberry gateware.

When Takashi, JI2UDD and I were working on the Hermes Lite 2 Plus, I found the same slow (but maybe not as slow as I am seeing with the Radioberry) RX to TX turn-around in CW operation.  Takashi made a number of changes to improve this and CW works very, very well with the Hermes Lite 2 Plus.  My recollection is that Takashi and Steve had some communication about these issues and Steve made at least one of the changes Takashi came up with.  I believe there was one or more that Steve was not able to reproduce/corroborate and so those changes were not made.  Again, this is my recollection.

I will probably (temporarily) reload the gateware in my Hermes Lite 2 Plus with more recent Hermes Lite 2 gateware to see if I can reproduce what I am seeing with the Radioberry (with QSK TR Switch and QSX PA) but wanted to post here to report my progress and findings.

P.S.  I also noticed that the slight difference in output between CW and Tune modes at the Radioberry TX output is more pronounced/noticeable with the power amplifier in place.  It is not necessarily a problem but I am wondering why CW is handled differently than Tune.  I do not know anything about the intricacies of the gateware in this regard but intuitively I would think that CW and Tune operation would in principal be the same thing.

73,

Robert, WA2T

[rentwist]

P.S.  I was wondering if there was a Full Duplex setting (to turn off) for the RX somewhere in PiHPSDR, I could not find one.  The TR switch isolation allows about -3 dBm TX at the RX port during transmit.  The RX recovery when reutnring from TX can be observed on the S-meter.  It takes ~1-2 seconds for the reading to return to the noise floor.

[pa3gsb]

Robert,

I did by hand an update of the schematic.

Please review the sketch.  Also understood well that the isolation is 43dB?

Will look into your other remarks later.

73 Johan

PA3GSB

[rentwist]

Hi Johan,

Your schematic changes are correct but D3, C4, L3 and C8 can be deleted, thereby eliminating the "AMP_IN" port as it is not needed/used in this application.

Yes, I am measuring ~43 dB isolation at a PA output of ~10W (after the QSK TR Switch).  The PA power is important because D5 and D6 are actually a voltage doubler, which takes a sample of the RF output and rectifies/doubles it to make a high voltage which is used to reverse bias D1 and D2 to increase TX/RX isolation during TX.  I measured 60V with a scope at C7.

Also, with drive set to make a power output of ~10W using Tune produces ~14W using CW (using same drive setting), about a 1.5 dB difference.  I have set the PA menu for 51 dB, this causes a 50% drive setting to produce ~10W (using Tune) with the QSX PA (on 40m which is where I have been testing).

I may use a BSP122 MOSFET instead of the 2N7002 on my boards as it is the same SOT-223 footprint.  The 2N7002 is a SOT-23 and was a bit challenging to put on a SOT-223 footprint!

I will edit the schematic to reflect the changes and post it here.

Lastly, I had many of these PCB's made when I did the design back in 2017.  I would be happy to provide these to people wanting them for a small cost.  I have had good luck sending small boards like these using a regular letter envelope so the postage (even international) is very reasonable.

If there is interest, I can draw up a revised BOM to go with the revised schematic.

73,

Robert, WA2T

[rentwist]

I might not be quite right on the unneeded component removal, L3 and C8 may still be required.  C4 and D3 can be removed but D3 would need to be jumped.  Will confirm but this is what it is looking like upon a second look.

[rentwist]

Hi again Group,

After speaking with Jim Veatch (the Hardrock-50 designer - where I lifted the basic design from), he explained - and I agreed - that the best thing to do is to leave the AMP_IN port components in place and simply terminate that port with a 50 ohm resistor.  There is a lot more going on in the circuit than just DC/RF decoupling for the diode biasing and so leaving things alone is the best way to ensure no problems with the circuit.

I will edit the finalized schematic and post it.

73,

Robert, WA2T

[rentwist]

Terminating the AMP_IN port in 50 ohms improves the TX/RX isolation by ~6 dB!  So ~49 dB.  Pretty happy with that for a solid state QSK TR Switch.

[rentwist]

Hi yet again Group!,

So, in the end, the schematic is very close to correct as it is.

The changes are:

Remove R5

Change R7 to 10K

Change Q1 to BSP122 (or 2N7002 if you do your own PCB or don't mind putting an SOT-23 on a SOT-223 pads on my PCB)  Do not substitute Q2 as it must handle hundreds of volts.

Terminate AMP_IN with a 50 ohm resistor

Connections to board:

12V to RY1 Pin 1.  This is the DIP relay inboard pad labeled "12V" on my PCB.

Ground.  I used the via near the DIP relay pads.  I used two ground wires (both will fit in the one via if you use sufficiently small wires).  One to go with the 12V power supply and one to go with the PTT connection from the Radioberry.

Radioberry CN301 pin 3 (PTT) to RY1 Pin 16.  This is the DIP relay outboard pad labeled "TX" on my PCB.

Radioberry RX port to RY1 pin O1.  This is the DIP relay pad opposite/inboard of the pad labeled "PA IN (RX)" on my PCB.  The labeled pad is actually NC, I scraped away ground plane on the PCB bottom and grounded this pad to use it for the coax shield.

PA Output (actually after LPF board - to be added!) to RY1 pin S1.  This is the DIP relay pad opposite/inboard of the pad labeled "AMP OUT" on my PCB.  The labeled pad is actually NC, I scraped away ground plane on the PCB bottom and grounded this pad to use it for the coax shield.

RY1 pin P1 to Antenna.  This is the pad opposite/inboard of the pad labeled "PA_OUT (Ant)" on my PCB.  The labeled pad is actually NC, I used a small wire jumper on the PCB bottom side to bring the inboard pad connection to the outboard (labeled) pad.  Then, I installed a PCB edge mount SMA connector with the center pin soldered to the pad.  For the ground pins of the connector, I scraped away ground plane on the PCB bottom and soldered the two SMA connector pins to the ground plane.

Here's a picture:

And a picture of my test setup (the N2ADR LPF board will be the next addition):

73,

Robert, WA2T

[rentwist]

Hi Group,

This morning I put together a BOM for the Solid-State QSK TR Switch, but not sure of the best way to share it.  There does not appear to be file storage on Google Groups.  If interest takes off, I will put board documentation up on my Githib page.

For now, I will email it to those who have purchased a board.

The cost for a QSK TR Switch PCB is $7.50 shipped in the USA.  Elsewhere, $7 plus the cost of shipping (which for an letter envelope will be very reasonable).  I have many of these boards available.  Please email me off forum with interest.

73,

Robert, WA2T

[pa3gsb]

Robert

Are you using the PTT also for switching on/off of the PA?

If the PTT is only attached to the switch ; you created a single point of failure.... if the PTT wire get loose during experiments the rf is visible at the RX entrance of the Radioberry.

I think it would be fine to have an additional RF sense circuit at the RB TX output (may be also from PA out) which will be there for protection.

73 Johan

PA3GSB

[Bob A. Booey]

Hi Johan,

With the QSK switch, there is always TX/RX isolation regardless of the switch state so no worries there.

73,

Robert

Sent from my iPhone

On May 31, 2020, at 6:34 AM, pa3gsb <pa3...@gmail.com> wrote:



--

You received this message because you are subscribed to the Google Groups "Radioberry" group.

To unsubscribe from this group and stop receiving emails from it, send an email to radioberry+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/radioberry/41d24bf5-02d5-4f75-82ec-9cc1ee88eda0%40googlegroups.com.

[rentwist]

I am not using PTT to control the PA.  The PA has 12V applied to it all the time.  I was considering bringing out the drain connection of Q1 on the QSK TR Switch board for PA bias control (so bias would only be present when TX is active).  Bias control is low active on the QSX PA.

[rentwist]

Hi Group,

Complete documentation .zip for the QSK TR Switch attached.

73,

Robert, WA2T

[Delroy L]

I am wondering which file do you use to order the PCB?

[rent...@comcast.net]

Hi Delroy (and group),

I have plenty of these boards available for sale but am away from home currently.  I will let everyone know when I am back home and able to send boards out.

73,

Robert, WA2T

[Kun Lin]

Thanks!

[Are Barstad]

Hi Robert,
I'd love a board if you have any left. I'm in Norway, but will of course pay for shipping.

Regards,

Are - LB3SA

[Joey G. (Bonehead)]

Hello.  Curious if you still had a board available for purchase.

Thanks 73

Joey G.

K5BHD

[rent...@comcast.net]

To be clear, what I have available are bare boards.

73,

Robert, WA2T

[Ian Clark]

Hi Robert,

Do you still have any bare pcb's left?

If so could you let me know how to order one delivered UK

Regards

Ian

[rent...@comcast.net]

Hi Group,

Please see www.wa2tdesigns.com for the QSK Board and a number of other things I make, mostly for (but not only!) SDR's.