I used my old nanovna for this so I was curious about frequency accuracy. I bought a "spare" nanovna that I have only had for a few days. It reads the same frequency for the peak on 10 meters, 28052. So, I imagine the readings are pretty close to correct.
Tony
AD0VC
toggle quoted message
Show quoted text
All right, so the mixer input impedance for the practical QMX BPF/matching network design purposes, we can agree that the mixer input is 75 to 120 ohm (maybe even higher on lower bands), a bit reactive. It's good to confirm my working assumption was reasonable.
|
Solder suckers are "less than optimal" for desoldering.? Wicking with a braid soaked with flux, works very well, if you have the patience to do it.? It is also the most safe way, provided the iron temperature is set right...? Wicking is especially good, if the "lands and vias" are tiny and very fragile...
Regards
wb7ond
|
Re: DRA818V as a CW Beacon ?
On Sat, Oct 21, 2023 at 11:26 AM, Da Amazin' man G0FTD wrote:
On Sat, Oct 21, 2023 at 07:21 PM, Hugh Duff wrote:
DRA818V VHF
Try another group.
This is for QRP LABS products.
?
--
- 73 de Andy -
Which uses DRA818V!
|
All right, so the mixer input impedance for the practical QMX BPF/matching network design purposes, we can agree that the mixer input is 75 to 120 ohm (maybe even higher on lower bands), a bit reactive. It's good to confirm my working assumption was reasonable.
|
Good for you! While testing my QCX-mini, I managed to short the board and blew out a couple of traces. This was my first build and I went to Hans, thinking I needed a new board. His suggestion was to simply solder wire to replace the traces. I got some 30ga Kynar wire and successfully made the repair. Excellent advice to a novice builder I¡¯m glad to hear your story. You don¡¯t learn as much from your successes as you do from overcoming problems and obstacles! -- Thanks and 73,
Tim
VE1XR
|
28.074MHz with 250Hz resolution.
Tony
AD0VC
toggle quoted message
Show quoted text
Thanks for your effort.
So, 10-12kHz away from the LO frequency the input impedance drops in half on 20m. So, in order not to complicate the IF/baseband signal processing to optimize for the best image rejection, I think it is best to keep the mixer input signal line to a reasonably
low level. We still get some of the preselector performance boost but not as much as possible with a direct conversion detector case.
|
Thanks for your effort.
So, 10-12kHz away from the LO frequency the input impedance drops in half on 20m. So, in order not to complicate the IF/baseband signal processing to optimize for the best image rejection, I think it is best to keep the mixer input signal line to a reasonably low level. We still get some of the preselector performance boost but not as much as possible with a direct conversion detector case.
|
toggle quoted message
Show quoted text
PITA. I have to recalibrate every time I change a frequency sweep setting. With 5 averages it looks better but takes forever to do.
I increased font size.?
14074 Zoomed, 250Hz resolution, 5 averages
Tony
AD0VC
Oh, ok. We were writing on the same thing simultaneously.
Can the trace be made smoother by increasing the number of averaging?
The impedance peak looks like two-tier structure and I am curious about those each peak. Hope that the broader peak is wide enough for ideal superhet and image rejection...
|
PITA. I have to recalibrate every time I change a frequency sweep setting. With 5 averages it looks better but takes forever to do.
I increased font size.?
14074 Zoomed, 250Hz resolution, 5 averages
Tony
AD0VC
toggle quoted message
Show quoted text
Oh, ok. We were writing on the same thing simultaneously.
Can the trace be made smoother by increasing the number of averaging?
The impedance peak looks like two-tier structure and I am curious about those each peak. Hope that the broader peak is wide enough for ideal superhet and image rejection...
|
Re: DRA818V as a CW Beacon ?
Hugh,
QRPLabs products include LightAPRS which uses this module.
If you enter 'DRA818V' in the search engine in the topic, you can see the comments related to the topic.
If you want to read about LightTracker, .
--
Gyula HA3HZ
|
Re: DRA818V as a CW Beacon ?
That sounds like an ideal transceiver for packet radio... though 25kHz channel separation.
If keying is sought without QSK capability, you could try dong that with a PIN diode attenuator so that the transmitted power is absorbed by designated resistor(s). The easiest is to use two diodes in a series-shunt config with a quarter-wavelength coax as the impedance inverter. Another way, which requires a few more components but wide band, is a T-network attenuator configuration. Either way, the attenuator is bypassed during receive.
I'd look into GMP4201-GM1, GMP4215-GM1, MPP4203, etc. for the diodes.
If you are going to build an amp for a bit more power, keying can be done in the drive and the final, as well.
|
Re: DRA818V as a CW Beacon ?
On Sat, Oct 21, 2023 at 07:21 PM, Hugh Duff wrote:
DRA818V VHF
Try another group. This is for QRP LABS products. ? -- - 73 de Andy -
|
Hello group.
I've read a bit of chatter on the DRA818V VHF transceiver module around here but I'm wondering if any of you have successfully used one of these as a 2m CW beacon ???The well known TX-RX turnaround delay seems to render the prospect of keying the PTT like a CW transmitter useless. Can this thing handle 100% duty cycle (I would use it in low power mode and heatsink it as necessary) and has anyone delved into the pcb to find a suitable place to key the TX chain without creating any chirp or other undesirable effects ??
BTW, does anyone have a link to a schematic for this module ?
Thanks, Hugh VA3TO
|
Oh, ok. We were writing on the same thing simultaneously.
Can the trace be made smoother by increasing the number of averaging?
The impedance peak looks like two-tier structure and I am curious about those each peak. Hope that the broader peak is wide enough for ideal superhet and image rejection...
|
Why do the peak impedances not agree between the big span shots and the zooms? Is that because the measurement frequency steps are different? Can the measurement span be narrowed and/or steps be made finer grained so that the two measurements agree?
|
I think those impedances are misleading because of the large vna step size. In this attachment I have dropped to a 500hz step and you can see that the response is much narrower and Z peaks higher.
Tony
AD0VC
toggle quoted message
Show quoted text
On Sat, Oct 21, 2023 at 01:21 PM, John Z wrote:
Any notion that the Tayloe input is a simple low impedance load is now dispelled.
I don't think anyone advanced such a notion in this group as far as I know. I also don't recall anyone who argued against that the input impedance is dependent on fLO.
About 120ohm on 20m and 38ohm on 10m is a lot lower than what was predicted from your simulation that omitted parasitics.
Also, keep in mind that QMX uses an IF of 12kHz so the actual operating impedance is much lower than those peak impedance right at fLO. Exactly how much, I'll wait for a non-blurry closeup traces.
|
On Sat, Oct 21, 2023 at 01:51 PM, John Z wrote:
Ok on your and Steve's past opinions. I wish that was indicated with the original statement.
I would also guess that the QDX and QMX series-LC BPF's were designed
with the idea of a low Tayloe input impedance in mind. When you
calculated your BPF values, what did you assume about the Tayloe
input?
I did a few experiments by shunting the mixer input with a few resistors and capacitors. I got an impression that the impedance (at the receive frequency) was between 75 and 150 ohm, slightly reactive. The load impedance only affects the filter Q and not the fc as long as the load is purely resistive, which it isn't. But a little adjustment to the C is enough to absorb that difference, if it is a problem at all (since the filter Q is low anyway). Anyway, I knew the actual preselector effect was better than the raw BPF response, and this thread just reinforces that view.
Another issue, which is actually just as significant but no one else is talking about is the stray shunt cap at the input of the BPF. That drags the line impedance down and capacitive on high bands.
|
Yes. I skipped a step. I had to redo the setup and ran into a crappy ground connection on my sma conector. Finally recaptured 14MHz zoom.
14.074MHz zoomed
toggle quoted message
Show quoted text
I think one of your zooms is mixed up (supposed to be 20m appears to be actually 10m) but I get the idea. More of an issue I have is the blurry images that makes it hard to read...
The input impedance appears to be about 120ohm up to 17m but goes down much lower at 15m and higher.
In order to take advantage of the input impedance characteristics to maximum advantage for the preselecter purpose is to keep the signal impedance high. Since 15m and higher need a lower input impedance than 17, this is probably most easily achieved by a
high-pass L-match (series C to the source, shunt L at the load). If done well, some 6dB sensitivity gain can be expected on lower bands, less gain on higher bands.
I said L-match above and in another thread since the explicitly added components look like an L network. However, keep in mind the input end has a significant shunt cap (BSS123, MUX, PCB pattern, etc.) so the actual function is more like a pi-match if you
start with the end of LPF output. What I did in my 60-10m QMX is a pi-match with the inductor in the middle. That serves the same impedance matching function, except it is a low-pass configuration (because I wanted to use the existing network topology with
minimum modifications).
|
I put together a QCX+ for 40 meters and just finished running through the preliminary alignment.
Lots of fun and a tribute to Hans' creativity. I did have one experience I would like to pass on.
Somehow, don't ask me how, I had put the three connectors on the front panel pcb in backwards.
I don't know how it happened. I blame gremlins who arrived late at night and attacked the board.
Anyway, I was going to order a new board and parts from Hans, but had second thoughts. I should
at least try to remove the connectors first before ditching the parts, pcb and building a new
unit. It seemed like a waste and I am cheap. I used a solder sucker and gentle pulling on the connectors
to loosen them and then pulled to top of the connectors off the board. Of course, the connectors
were destroyed during this surgery, but I soon found they are cheap. I then cleaned the holes in the pcb
with the solder sucker so I could see daylight on the other side looking through the holes in the pcb.
I put new connectors in and? and proceeded with testing. I could not navigate the command menu
using the optical encoder so I knew I had probably damaged a trace(s) removing the connectors. So
I used the diagram of the front panel pcb traces on page 12 of the manual and a VOM to test for
continuity of the traces shown on the diagram. I found the trace from the optical encoder to the middle
connector to be open. I soldered a jumper to reestablish continuity from the encoder to the connector
and all was well. In hindsight, I should have just crushed the connectors and then? gently removed the
prongs one by one then replaced each connector with a new one. Don't get old!
|
Ryuji,
Both Steve K1RF and I once held that idea and we expressed it here.
Steve had the great idea to take the question to Dan Tayloe directly.
Dan's answer was that at the clock frequency the input looks like an
open circuit, and falls away to very low values as separation between
input and clock increases. You can find all that in another thread on
the forum.
Dan's answer was of course predicated on a lack of parasitics, as is
my simulation. Tony's real world results are validating as well,
although we don't know what peak values might have been obtained if
the NanoVNA hit the LO frequency precisely.
I would also guess that the QDX and QMX series-LC BPF's were designed
with the idea of a low Tayloe input impedance in mind. When you
calculated your BPF values, what did you assume about the Tayloe
input?
JZ
toggle quoted message
Show quoted text
On Sat, Oct 21, 2023 at 1:32?PM Ryuji Suzuki AB1WX <ab1wx@...> wrote:
On Sat, Oct 21, 2023 at 01:21 PM, John Z wrote:
Any notion that the Tayloe input is a simple low impedance load is now dispelled.
I don't think anyone advanced such a notion in this group as far as I know. I also don't recall anyone who argued against that the input impedance is dependent on fLO.
About 120ohm on 20m and 38ohm on 10m is a lot lower than what was predicted from your simulation that omitted parasitics.
Also, keep in mind that QMX uses an IF of 12kHz so the actual operating impedance is much lower than those peak impedance right at fLO. Exactly how much, I'll wait for a non-blurry closeup traces.
|
Barb WB2CBA
Ryuji Suzuki AB1WX
Tim Delaney
HA3HZ