Re: Urban HF Antenna Dilemma: LZ1AQ vs. M0AYF
Caaarlo,
Would you publish as attachments complete s1p files for both amplifier and your loop antenna ?
I am planning to check whether an impedance matching network can be synthesized for this broadband 1- 30 MHz application.
Regards,
Raphael
|
The output transformer is now the limiting factor for 30 MHz high
end gain.
The 2SC5551 is much less susceptible to UHF oscillation than many
other such transistors, though some circuits using it still
require ferrite beads to tame oscillations. The LZ1AQ circuit does
not need them.
73,
Steve AA7U
On 7/12/2024 2:56 PM, W0LEV via
groups.io wrote:
toggle quoted message
Show quoted text
The
ft of the 2SC5551 is 3.5 GHz.? This
should show absolutely no gain slump at 30 MHz.? But,being in
the GHz range, it will be far more difficult to "tame".? Good
?W layout techniques will be required.
As
for the size of the NanoVNA and the 1:1 "balun", yes, the Nano
is small enough that the balun is not necessary at HF
frequencies so long as short leads are used in the test
setup.? I can not use my 8753C for such measurements!! ??
Dave
- W?LEV
|
The ft of the 2SC5551 is 3.5 GHz.? This should show absolutely no gain slump at 30 MHz.? But,being in the GHz range, it will be far more difficult to "tame".? Good ?W layout techniques will be required.
As for the size of the NanoVNA and the 1:1 "balun", yes, the Nano is small enough that the balun is not necessary at HF frequencies so long as short leads are used in the test setup.? I can not use my 8753C for such measurements!! ??
Dave - W?LEV
|
Re: Steve ugly-build LZ1AQ using MPS2222 sweeps uploaded to Photos (revised!)
Look up "gain-bandwidth product".?? This amp has about +25 dB
gain. If it had 0 dB gain then it could work at its rated max
frequency.
73,
Steve AA7U
On 7/12/2024 2:49 PM, W0LEV via
groups.io wrote:
toggle quoted message
Show quoted text
The
ft of the '2222 is 250 to 300 MHz.? How
could that account for gain slump at 30 MHz?
Dave
- W?LEV
Thanks to Caaarlo for pointing out my long input lead
problem. I've redone the MPS2222 build plus built a second
one using 2SC5551. I've deleted the old picture and sweeps
from the folder and put new pictures showing the short
input leads and the input coupling transformer with the
shortest leads I could use. I've added a lot of info to
the folder description so be sure to click on it so it all
shows up. Reducing lead length indeed lowers the input
impedance at the 30 MHz high end, but has no effect on
loop amp actual gain. It's interesting that the MPS2222
has a slightly lower input impedance than the 2SC5551
version. (The same calibrated-out input transformer was
used for both loop amp sweeps.)
/g/loopantennas/album?id=296351
73,
Steve AA7U
|
Re: Steve ugly-build LZ1AQ using MPS2222 sweeps uploaded to Photos
Ok, I stand corrected, as I was doing things wrong, both maths and sweep:
1-The VNA was being saturated with 10dB above its dynamic range.
2-THe mismatch attenuation calculations were wrong, so the input level to LZ1AQ is actually around -20dBm
After inserting a 25dB attenuator on the output of LZ1AQ, I got the following sweep:
/g/loopantennas/photo/296346/3804920
So, Steve, you were right. There is some evident gain oll off, but still, those long capacitor leads are worsening it. It might be worth shortening the input path.
The corrected gain values, adding the mismatch attenuation factor is as follows:
1MHz
S21=25,855dB
Mismatch attenuation = 7,051dB
Gain = 32.906dB (S21 + Mismatch attenuation)
?
3.61MHz
S21 = 25,867dB
Mismatch attenuation = 6.973dB
Gain = 32.84dB
7.09MHz
S21 = 25.607dB
Mismatch attenuation = 6.934dB
Gain = 32.541dB
14.34MHz
S21= 24.468dB
Mismatch Attenuation = 6.604dB
Gain = 31.072
30MHz
S21=21.345dB
Mismatch Attenuation = 4.14
Gain = 25.485dB
I am not happy with the -20dbm at LZ1AQ input, as this leads to almost 13 dbm output at 1MHZ which might be reaching the compression region. I might try inserting an attenuator on the input and compare results.
73
|
Re: Steve ugly-build LZ1AQ using MPS2222 sweeps uploaded to Photos (revised!)
The ft of the '2222 is 250 to 300 MHz.? How could that account for gain slump at 30 MHz?
Dave - W?LEV
Thanks to Caaarlo for pointing out my long input lead problem.
I've redone the MPS2222 build plus built a second one using
2SC5551. I've deleted the old picture and sweeps from the folder
and put new pictures showing the short input leads and the input
coupling transformer with the shortest leads I could use. I've
added a lot of info to the folder description so be sure to click
on it so it all shows up. Reducing lead length indeed lowers the
input impedance at the 30 MHz high end, but has no effect on loop
amp actual gain. It's interesting that the MPS2222 has a slightly
lower input impedance than the 2SC5551 version. (The same
calibrated-out input transformer was used for both loop amp
sweeps.)
/g/loopantennas/album?id=296351
73,
Steve AA7U
On 7/11/2024 8:14 PM, Caaarlo via
wrote:
On Thu, Jul 11, 2024 at 09:41 PM, Steve Ratzlaff wrote:
The
transistors run out of gain-bandwidth product at 30 MHz as you
can see
from the sweep (the gain is rolling off). Caaarlo's loop amp
should have
similar gain since both used the same transistors.
Steve,
Thank you for the compliments.
I have just measured gain as per your request, and mine is
almost flat:
1MHz 33.15dB
7.1MHz 33.51dB
30MHz 33.50dB
At 1MHz I am sure that the lower gain is due to 1uF input
capacitors used, which add some reactance hence reducing input
level, output level and measured gain. Some 10 or 100uF should
lift the gain to the same level as the other frequencies.
Regarding your gain rolling off, it is not a problem of
transistors, but a problem of ugly-build as you just named it.
For being such a low impedance input, those input capacitors,
their legs, and everything in between your generator and the
emitters,? are inserting an important amount of reactance in
series with the actual input of the amplifier: the emitters.
So you basically have a voltage divider consisting of a large
parasitic inductance and the smaller input impedance, reducing
the actual power delivered to the input hence reducing output.
Parasitic inductance is not a problem when load impedances are
much higher, but when they are as low as in LZ1AQ amp, in the
order of 2-4 ohms, they become a serious threat to performance.
A quick calculation:
those extended legs are probably 25mm for each capacitor, total
50mm, that is an inductance of about 52nH
At 1MHz the reactance is 0.3 ohms
At 30MHz it skyrockets to 9 ohms, which is at least twice the
resistive component at the emitters.
I would strongly recommend to shorten all input paths as much as
possible, because what you are experiencing has nothing to do
with the transistors.?
73
|
The following photos have been
uploaded to the
Caaarlo LZ1AQ sweep
photo album of the loopantennas@groups.io group.
By:
Caaarlo <caaarlo@...>
|
Re: Steve ugly-build LZ1AQ using MPS2222 sweeps uploaded to Photos (revised!)
Thanks to Caaarlo for pointing out my long input lead problem.
I've redone the MPS2222 build plus built a second one using
2SC5551. I've deleted the old picture and sweeps from the folder
and put new pictures showing the short input leads and the input
coupling transformer with the shortest leads I could use. I've
added a lot of info to the folder description so be sure to click
on it so it all shows up. Reducing lead length indeed lowers the
input impedance at the 30 MHz high end, but has no effect on loop
amp actual gain. It's interesting that the MPS2222 has a slightly
lower input impedance than the 2SC5551 version. (The same
calibrated-out input transformer was used for both loop amp
sweeps.)
/g/loopantennas/album?id=296351
73,
Steve AA7U
On 7/11/2024 8:14 PM, Caaarlo via
groups.io wrote:
toggle quoted message
Show quoted text
On Thu, Jul 11, 2024 at 09:41 PM, Steve Ratzlaff wrote:
The
transistors run out of gain-bandwidth product at 30 MHz as you
can see
from the sweep (the gain is rolling off). Caaarlo's loop amp
should have
similar gain since both used the same transistors.
Steve,
Thank you for the compliments.
I have just measured gain as per your request, and mine is
almost flat:
1MHz 33.15dB
7.1MHz 33.51dB
30MHz 33.50dB
At 1MHz I am sure that the lower gain is due to 1uF input
capacitors used, which add some reactance hence reducing input
level, output level and measured gain. Some 10 or 100uF should
lift the gain to the same level as the other frequencies.
Regarding your gain rolling off, it is not a problem of
transistors, but a problem of ugly-build as you just named it.
For being such a low impedance input, those input capacitors,
their legs, and everything in between your generator and the
emitters,? are inserting an important amount of reactance in
series with the actual input of the amplifier: the emitters.
So you basically have a voltage divider consisting of a large
parasitic inductance and the smaller input impedance, reducing
the actual power delivered to the input hence reducing output.
Parasitic inductance is not a problem when load impedances are
much higher, but when they are as low as in LZ1AQ amp, in the
order of 2-4 ohms, they become a serious threat to performance.
A quick calculation:
those extended legs are probably 25mm for each capacitor, total
50mm, that is an inductance of about 52nH
At 1MHz the reactance is 0.3 ohms
At 30MHz it skyrockets to 9 ohms, which is at least twice the
resistive component at the emitters.
I would strongly recommend to shorten all input paths as much as
possible, because what you are experiencing has nothing to do
with the transistors.?
73
|
Re: Steve ugly-build LZ1AQ using MPS2222 sweeps uploaded to Photos
On Fri, Jul 12, 2024 at 08:41 PM, Everett N4CY wrote:
When Sweeping a LZ1AQ amp you need to use a 1:1 transformer between the VNA and the amplifier.
Ideally, but when using a stand alone VNA, that is not connected to anything, it is so small in terms of wavelength, that it is effectively "floating" and balanced without the need for an extra transformer. This is the big advantage of the battery powered, pocket sized VNA's. You don't need to worry about the influence of the test gear adding stray circuit paths. But you do need to make the input Z and gain measurements separately, to avoid unbalancing the input during the Z measurements. If the output is connected to VNA port 2, it will ground one of the amplifier inputs. As an experiment, I just tried adding a 1:1 transformer on the input, and it made practically no difference to the measurements. Regards, Martin
|
Re: Steve ugly-build LZ1AQ using MPS2222 sweeps uploaded to Photos
On Fri, Jul 12, 2024 at 08:27 PM, W0LEV wrote:
I have no idea of the IP1 (1 dB output compression power).
Mine is around +23dBm I also tried varying the sweep power, no significant changes noticed up to the maximum level of 0dBm. Regards, Martin
|
Re: Steve ugly-build LZ1AQ using MPS2222 sweeps uploaded to Photos
When Sweeping a LZ1AQ amp you need to use a 1:1 transformer between the VNA and the amplifier. You also need to calibrate the VNA on the amp side of the 1:1 transformer. What I use for a 1:1 transformer is a BN73-302 binocular core with 6 turns on each side.
?
Steve and I both use the SDR-KITS VNA, which does very good job.
?
Everett N4CY?
In a message dated 7/12/2024 3:27:41 PM Central Daylight Time, davearea51a@... writes:
?
I'd also suggest -50 or -60 dBm input through the CM to DM transformer.? With gain, -30 dBm is a bit much to assure remaining in the linear region.? Used as the '2222 are in the amp, I have no idea of the IP1 (1 dB output compression power).
?
And, yes, absolutely use a CM to DM 1:1 transformer on the input.??
?
Dave - W?LEV
While replying to steve I realized I might be close to saturating the VNA input. So I will insert an attenuator after LZ1AQ output and see how it goes.
?
?
--
|
Re: Steve ugly-build LZ1AQ using MPS2222 sweeps uploaded to Photos
I'd also suggest -50 or -60 dBm input through the CM to DM transformer.? With gain, -30 dBm is a bit much to assure remaining in the linear region.? Used as the '2222 are in the amp, I have no idea of the IP1 (1 dB output compression power).
And, yes, absolutely use a CM to DM 1:1 transformer on the input.??
Dave - W?LEV
While replying to steve I realized I might be close to saturating the VNA input. So I will insert an attenuator after LZ1AQ output and see how it goes.
|
Re: Steve ugly-build LZ1AQ using MPS2222 sweeps uploaded to Photos
On Fri, Jul 12, 2024 at 06:29 PM, Caaarlo wrote:
I will insert an attenuator after LZ1AQ output and see how it goes.
I just found an old LZ1AQ board I prototyped some time ago using MP222A transistors. I connected a 20 dB attenuator on the amplifier output, and made a quick 1-50MHz sweep using a LiteVNA64 and internal storage. /g/loopantennas/photo/267767/3804910Regards, Martin
|
Re: Steve ugly-build LZ1AQ using MPS2222 sweeps uploaded to Photos
While replying to steve I realized I might be close to saturating the VNA input. So I will insert an attenuator after LZ1AQ output and see how it goes.
|
Re: Steve ugly-build LZ1AQ using MPS2222 sweeps uploaded to Photos
On Fri, Jul 12, 2024 at 12:25 AM, Steve Ratzlaff wrote:
Hi Caaarlo,
Thanks for your interesting comments. Please tell me the exact setup and equipment you used to do your own gain measurements, and what input level did you use?
Thanks,
Steve AA7U
Steve, My setup is simple: The VNA is a NanoVNA, which is a self-powered VNA, hence no ground influence, same goes for the laptop connected to it. 1) I did not use any balun or transformer to couple the VNA to the balanced input of the LZ1AQ.
2) The output of the LZ1AQ was also connected directly to the VNA3) The actual power entering the amplifier is around -30dBm.
4) The turns ratio of the trifilar output transformer is 6-6-5, on a small unknown toroidal core of high ? (Mix 73 or 75).Some words explaining each point of my setup: 1- No input transformer: First, all the setup is floating (no ground). Second, the input impedance is so low and dominant compared to the possible external couplings and interactions with the generator, the computer and ground (capacitive and/or inductive) that it ensures symmetry by itself. Adding a transformer is not a free lunch. One has to evaluate whether the balance of pros and cons is beneficial. 2- The output of the LZ1AQ was also connected directly to the VNA, as the output coil is floating, and could be theoretically used with either a balanced or unbalanced load. In reality there is some capacitive coupling of about 20pF to ground from the primary windings that might slightly alter the "floating condition" of the VNA so perhaps a current balun would have optimized the output measurement. But 20pF have a reactance of 360 ohms at 30MHz, more than 2 orders of magnitude higher compared to 1/2 Zin (~3 ohms) that connects the ground of the vna to the ground of LZ1AQ. The influence is minimal. So. as you can see, I consider that the ill effects of adding transformers, in this case, would outweight the light ill effects of not using them. 3) Input level: The NanoVNA has a fixed level and it is -13dBm within the HF range. But due to the huge mismatch presented by the very low Zin of LZ1AQ, the actual power entering the amplifier is around -30dBm.This reduction of around 18dB (depends on Zin at each frequency) has to be considered when computing gain.? I have to do it manualy, as the NanoVNA software is quite basic. 4) The turns ratio of the trifilar output transformer is 6-6-5, on a small toroidal core of high ? (Mix 73 or 75) theoretically optimal for 69 ohms, but either 75 or 50 ohm loads can be connected. 73 !!
|
Re: Steve ugly-build LZ1AQ using MPS2222 sweeps uploaded to Photos
|
Re: Steve ugly-build LZ1AQ using MPS2222 sweeps uploaded to Photos
On Fri, Jul 12, 2024 at 03:14 AM, Caaarlo wrote:
I would strongly recommend to shorten all input paths as much as possible, because what you are experiencing has nothing to do with the transistors.?
73
Hi Steve and Caaarlo, As an aside. If you use the same construction technique as shown in the photo that Caaarlo posted. /g/loopantennas/photo/296346/3804668Initially cover the whole board in self-adhesive Kapton, and then just cut out areas where you need a ground connection with a scalpel, you will be able to achieve much greater circuit density, and make prototypes a lot quicker than by laying down insulation where it is needed. This is now my preferred method of prototyping, and I also use thin brass sheet (from model suppliers) instead of PCB material, usually when I need improved heat sinking. Regards, Martin
|
Re: Steve ugly-build LZ1AQ using MPS2222 sweeps uploaded to Photos
Hi Caaarlo,
Thanks for your interesting comments. Please tell me the exact
setup and equipment you used to do your own gain measurements, and
what input level did you use?
Thanks,
Steve AA7U
On 7/11/2024 8:14 PM, Caaarlo via
groups.io wrote:
toggle quoted message
Show quoted text
On Thu, Jul 11, 2024 at 09:41 PM, Steve Ratzlaff wrote:
The
transistors run out of gain-bandwidth product at 30 MHz as you
can see
from the sweep (the gain is rolling off). Caaarlo's loop amp
should have
similar gain since both used the same transistors.
Steve,
Thank you for the compliments.
I have just measured gain as per your request, and mine is
almost flat:
1MHz 33.15dB
7.1MHz 33.51dB
30MHz 33.50dB
At 1MHz I am sure that the lower gain is due to 1uF input
capacitors used, which add some reactance hence reducing input
level, output level and measured gain. Some 10 or 100uF should
lift the gain to the same level as the other frequencies.
Regarding your gain rolling off, it is not a problem of
transistors, but a problem of ugly-build as you just named it.
For being such a low impedance input, those input capacitors,
their legs, and everything in between your generator and the
emitters,? are inserting an important amount of reactance in
series with the actual input of the amplifier: the emitters.
So you basically have a voltage divider consisting of a large
parasitic inductance and the smaller input impedance, reducing
the actual power delivered to the input hence reducing output.
Parasitic inductance is not a problem when load impedances are
much higher, but when they are as low as in LZ1AQ amp, in the
order of 2-4 ohms, they become a serious threat to performance.
A quick calculation:
those extended legs are probably 25mm for each capacitor, total
50mm, that is an inductance of about 52nH
At 1MHz the reactance is 0.3 ohms
At 30MHz it skyrockets to 9 ohms, which is at least twice the
resistive component at the emitters.
I would strongly recommend to shorten all input paths as much as
possible, because what you are experiencing has nothing to do
with the transistors.?
73
|
Re: Steve ugly-build LZ1AQ using MPS2222 sweeps uploaded to Photos
On Thu, Jul 11, 2024 at 09:41 PM, Steve Ratzlaff wrote:
The
transistors run out of gain-bandwidth product at 30 MHz as you can see
from the sweep (the gain is rolling off). Caaarlo's loop amp should have
similar gain since both used the same transistors.
Steve,
Thank you for the compliments.
I have just measured gain as per your request, and mine is almost flat:
1MHz 33.15dB
7.1MHz 33.51dB
30MHz 33.50dB
At 1MHz I am sure that the lower gain is due to 1uF input capacitors used, which add some reactance hence reducing input level, output level and measured gain. Some 10 or 100uF should lift the gain to the same level as the other frequencies.
Regarding your gain rolling off, it is not a problem of transistors, but a problem of ugly-build as you just named it. For being such a low impedance input, those input capacitors, their legs, and everything in between your generator and the emitters,? are inserting an important amount of reactance in series with the actual input of the amplifier: the emitters.
So you basically have a voltage divider consisting of a large parasitic inductance and the smaller input impedance, reducing the actual power delivered to the input hence reducing output.
Parasitic inductance is not a problem when load impedances are much higher, but when they are as low as in LZ1AQ amp, in the order of 2-4 ohms, they become a serious threat to performance.
A quick calculation:
those extended legs are probably 25mm for each capacitor, total 50mm, that is an inductance of about 52nH
At 1MHz the reactance is 0.3 ohms
At 30MHz it skyrockets to 9 ohms, which is at least twice the resistive component at the emitters.
I would strongly recommend to shorten all input paths as much as possible, because what you are experiencing has nothing to do with the transistors.?
73
|
Steve ugly-build LZ1AQ using MPS2222 sweeps uploaded to Photos
I had some MPS2222 transistors and built up a quick LZ1AQ and swept it, with the sweeps and picture of the ugly-build added to the Photos. This is only for comparing? with Caaarlo's build and sweeps (thanks, Caaarlo, you did a nice job building yours). His nanovna sweep didn't include gain sweep. I used the DG8SAQ VNWA with the same two frequency sweeps, and with calibrated-out 1:1 input transformer (unbalanced to balanced) which shows both input impedance and gain on the same sweep. I used -30 dBm input level to ensure the loop amp is not overloaded. I'm not sure what transformer and turns Caaarlo used; I used a BN73-302 binocular core with 8/8:6 turns which is what I use with a nominal 50 ohm coax cable output. (Same as what Everett Sharp N4CY's LZ1AQ PCB uses.) The transistors run out of gain-bandwidth product at 30 MHz as you can see from the sweep (the gain is rolling off). Caaarlo's loop amp should have similar gain since both used the same transistors. /g/loopantennas/album?id=296351??? If that doesn't work, search Photos for "Steve LZ1AQ Sweep".
|
Caaarlo
Everett N4CY