T3 Winding
316 views
Skip to first unread message
Dieter Kedrowitsch
May 22, 2018, 12:47:31 PM5/22/18
to Hermes-Lite
My Hermes Lite 2 has been running almost non-stop on 20M WSPR for over a week now without a single hiccup.
The only issue I'd like to look into is I've noticed my TX output is below 5 watts on many bands despite turning the PA gain to maximum. I'm using an N2ADR LPF board and when bypassing that, on many of the bands it makes the difference of say only 4.3 watts maximum vs 5.5 watts with the LPF bypassed. I imagine this insertion loss in the LPF is normal since the power reduction is simply the IMD products being reflected back to the finals.
I'd like to be certain I wound T3 properly. I used the recommended B62152A4X30 core and wound what I believe to be 1+1:4.
For the primary, using 22AWG stranded PTFE wire, I fed a Y-shaped center tapped wire one leg into each hole, then fed each leg back down it's opposite hole.
For the secondary, using the same 22AWG stranded PTFE wire, I did 4 U shaped turns--each consisting of the going in one hole and back down the other.
Does this sound correct?
What are the maximum power readings I should be expecting, with and without the LPF?
Cheers!
73 de W3DRK
The only issue I'd like to look into is I've noticed my TX output is below 5 watts on many bands despite turning the PA gain to maximum. I'm using an N2ADR LPF board and when bypassing that, on many of the bands it makes the difference of say only 4.3 watts maximum vs 5.5 watts with the LPF bypassed. I imagine this insertion loss in the LPF is normal since the power reduction is simply the IMD products being reflected back to the finals.
I'd like to be certain I wound T3 properly. I used the recommended B62152A4X30 core and wound what I believe to be 1+1:4.
For the primary, using 22AWG stranded PTFE wire, I fed a Y-shaped center tapped wire one leg into each hole, then fed each leg back down it's opposite hole.
For the secondary, using the same 22AWG stranded PTFE wire, I did 4 U shaped turns--each consisting of the going in one hole and back down the other.
Does this sound correct?
What are the maximum power readings I should be expecting, with and without the LPF?
Cheers!
73 de W3DRK
Steve Haynal
May 23, 2018, 12:50:29 AM5/23/18
to Hermes-Lite
Hi Dieter,
I'm not sure I follow your Y-shaped wire, but I'd describe the 1+1 primary as two pieces of wire, each goes through the left hole and out the right. Two ends are soldered in the center hole, pin 2 from this wiki page:
Single ends are soldered into pins 4 and 5 respectively.
The secondary is 4 turns. In one hole and out the other always counts as 1 turn.
You can find the expected PA output after the N2ADR filter board here:
For me, I was just under 5W on only 15 and 12M.
Maybe you don't have the filter band select codes setup properly in PowerSDR. Please try and compare with Quisk or SparkSDR. There is a YouTube video linked to from the wiki describing Quisk filter band setup. Jim has a picture on his web page.
73,
Steve
KF7O
Dieter Kedrowitsch
May 23, 2018, 8:53:04 AM5/23/18
to Hermes-Lite
Good morning Steve,
Yep, I have also tested in Quisk I'm certain my LPF bank selection settings are correct. When assembling my board I was careful to use the specified capacitor values in NPO/COG for the TX stages, so naturally I'm suspecting T3.
I attached a picture of what I mean about the Y-shaped primary winding I used, it was posted by another HL builder in this group so I assumed that was the method to use. It's interesting there's apparently two variations of T3 winding in use.
Instead of both primary wires being fed up one half and down the other in a hairpin fashion, the center tap is centered under the core and each half fed into it's own hole so the winding is shaped like a Y, then both wires fed back down the opposite holes so it's nice and neat with the center tap right in the middle where it's needed by the footprint.
Perhaps this winding method has a negative effect and would explain my low TX power? I have a spare core so I can easily wind another using the the method you describe to see if it makes a difference.
I will try and report back with my findings, but in the meantime I'm curious to know if anyone else has their T3 wound like I pictured and if so what their TX power is?
73
W3DRK - Dieter
Yep, I have also tested in Quisk I'm certain my LPF bank selection settings are correct. When assembling my board I was careful to use the specified capacitor values in NPO/COG for the TX stages, so naturally I'm suspecting T3.
I attached a picture of what I mean about the Y-shaped primary winding I used, it was posted by another HL builder in this group so I assumed that was the method to use. It's interesting there's apparently two variations of T3 winding in use.
Instead of both primary wires being fed up one half and down the other in a hairpin fashion, the center tap is centered under the core and each half fed into it's own hole so the winding is shaped like a Y, then both wires fed back down the opposite holes so it's nice and neat with the center tap right in the middle where it's needed by the footprint.
Perhaps this winding method has a negative effect and would explain my low TX power? I have a spare core so I can easily wind another using the the method you describe to see if it makes a difference.
I will try and report back with my findings, but in the meantime I'm curious to know if anyone else has their T3 wound like I pictured and if so what their TX power is?
73
W3DRK - Dieter
Steve Haynal
May 23, 2018, 12:14:10 PM5/23/18
to Hermes-Lite
Hi Dieter,
I think if your winding was off then the power output would be much lower, more like half of what is expected.
How are you measuring power? The N2ADR board power measurement is best with low SWR. You said you were bypassing the N2ADR board? All the measurements I make are into a dummy load. Do you have a scope and dummy load to measure your power that way? You might also want to check the low power TX output. It may be that your drive is low. Variations in capacitors in the balanced filter have led to low drive power in the past.
73,
Steve
KF7O
Dieter Kedrowitsch
May 23, 2018, 12:44:03 PM5/23/18
to Hermes-Lite
To measure my output power I've been transmitting into a dummy load while measuring with an LP-100A. The N2ADR power measurement don't seem to work in PowerSDR so I'll retest that with Quisk. I certainly can make some measurements with a scope connected to the dummy load as well.
And to bypass the LPF I simply cleared the J16 TX registers to prevent any of the LPF banks from being placed into circuit during transmit.
What would you consider to be a normal low power output?
73
Dieter
And to bypass the LPF I simply cleared the J16 TX registers to prevent any of the LPF banks from being placed into circuit during transmit.
What would you consider to be a normal low power output?
73
Dieter
in3otd
May 23, 2018, 1:04:02 PM5/23/18
to Hermes-Lite
Hello,
the TX low-power output should be between 16 dBm and 17 dBm, depending on the frequency; in this old post the green line in the first graph shows what I measured on a modified H-Lv3b3, which has practically the same circuit as the latest builds.
The first graph here shows what you should get at the PA output, without the filter board.
73 de Claudio, IN3OTD / DK1CG
the TX low-power output should be between 16 dBm and 17 dBm, depending on the frequency; in this old post the green line in the first graph shows what I measured on a modified H-Lv3b3, which has practically the same circuit as the latest builds.
The first graph here shows what you should get at the PA output, without the filter board.
73 de Claudio, IN3OTD / DK1CG
Steve Haynal
May 23, 2018, 1:06:17 PM5/23/18
to Hermes-Lite
Hi Dieter,
I think Claudio is pointing to past measurements which are also linked to from the performance and testing wiki page:
73,
Steve
KF7O
in3otd
May 23, 2018, 3:53:42 PM5/23/18
to Hermes-Lite
yep, I forgot we have that on the Wiki also, hi.
Dieter, from your description the transformer winding seems correct to me; anyway here is a rough drawing which tries to show how it's usually done
The secondary winding is either done with a "Y" or two "hairpins" connected together, the end result is the same. Polarity between primary and secondary is not important, so you can swap pins 4 and 5 on the primary and 1 and 3 on the secondary.
Check that the transformer core is not touching some pads on the PCB or a bare wire edge, as the core is conductive and other builders in the past saw degraded performances in this case. Also check if the VPA supply stays at 8 V with the PA at maximum power.
Dieter, from your description the transformer winding seems correct to me; anyway here is a rough drawing which tries to show how it's usually done
The secondary winding is either done with a "Y" or two "hairpins" connected together, the end result is the same. Polarity between primary and secondary is not important, so you can swap pins 4 and 5 on the primary and 1 and 3 on the secondary.
Check that the transformer core is not touching some pads on the PCB or a bare wire edge, as the core is conductive and other builders in the past saw degraded performances in this case. Also check if the VPA supply stays at 8 V with the PA at maximum power.
73 de Claudio, IN3OTD / DK1CG
Dieter Kedrowitsch
May 31, 2018, 9:47:38 PM5/31/18
to Hermes-Lite
Well I finally had a little time tonight to check the low-power and PA TX outputs.
I dragged out my spectrum analyzer and one of my scopes to take some quick measurements. From the low-power port, according to the SA I seem to get between 15.25-16 dBm on all bands. So while that sounds about a dBm lower than specified above, it doesn't seem too bad.
As for the PA output, I only measured with my scope since I need to dig up my RF sampler. When transmitting into a 50ohm dummy load, on 40 meters (the band with the lowest output) I recorded about 44.4Vpp with the LPF disabled, and 40.8Vpp with the LPF enabled. So about 37dBm without LPF, and 36dBm with, which is pretty close to what my trusty LP-100A is displaying. Other bands are about a dBm higher.
This seems to be about a dBm or two lower than what others seem to report from their HL2...
I'm about to finish building a 2nd Hermes Lite 2, so I'll be curious to compare measurements between the two of them...
Steve Haynal
Jun 1, 2018, 6:20:12 PM6/1/18
to Hermes-Lite
Hi Dieter,
Thanks for the update. This looks pretty close to what is expected. Just curious, did you substitute any values or manufacturers for parts in the TX preamp, balanced filter and on the N2ADR board?
If you'd like to increase your output a bit, you can increase the gain of the TX preamp. Originally R55 was 75 Ohms. This produced about 20 dBm but would often over drive the PA. We switched to the currently specified 120 Ohms and see around 17 dBm now. You may want to try a 100 Ohm resistor here.
For those interested in "over clocking" their HL2 for more RF output power, R55 and the voltage divider setting the PA supply voltage to 8V (R19 and R46) are components to tweak. The PA has been running cool so it should be possible to squeeze a few more watts out. You'd have to keep an eye on distortion though.
73,
Steve
KF7O
Amogh Desai
Jun 2, 2018, 5:36:03 AM6/2/18
to Hermes-Lite
Hi Steve,
I have a BN43-202, do i have to use a ptfe wire? Or a normal shielded wire will do? Also should be 22 or 24 gauge?
Amogh
VU3DES
Steve Haynal
Jun 3, 2018, 12:51:47 AM6/3/18
to Hermes-Lite
Hi Amogh,
I used AWG #22 enamel insulated wire with my BN43-202. The alternate smaller core will heat and also electrical conduct. These problems have not been seen with the BN43-202 core.
73,
Steve
KF7O
Dieter Kedrowitsch
Jun 5, 2018, 10:44:34 AM6/5/18
to Hermes-Lite
Good morning Steve!
I actually did have to substitute a few of the capacitors.
For B1-B122 I used GCJ188R71H104KA12D (https://www.digikey.com/product-detail/en/murata-electronics-north-america/GCJ188R71H104KA12D/490-5798-1-ND/2783823)
For C46, C89, C144, C145, C146, C147 I used CL21B105KAFNNNG (https://www.digikey.com/product-detail/en/samsung-electro-mechanics/CL21B105KAFNNNG/1276-6469-1-ND/5958097)
For C53 I used CC0805JRNPOABN100 (https://www.digikey.com/product-detail/en/yageo/CC0805JRNPOABN100/311-4245-1-ND/8025334)
So the only one of those substitutes in the pre-amp and PA RF chain is the .1uF caps where used as a coupling/DC blocking device. But in this role it's RF current and Q capabilities should be far less critical critical then if it were in a tuned circuit. I don't think it would be worth swapping those out, do you?
Also, I just finished up the second Hermes Lite for my brother last night. I didn't have enough of the 22AWG PTFE wire on hand so I wound T3 with 24AWG PCV insulated wire instead. The power output is even lower on this board, barely over 3 watts on all bands except around 14mhz. I'm grabbing more 22AWG wire today to rewind T3, hopefully that's the reason. I think I have an extra BN43-202 core laying around here somewhere, I may try switching to that core too.
Cheers!
Dieter - W3DRK
I actually did have to substitute a few of the capacitors.
For B1-B122 I used GCJ188R71H104KA12D (https://www.digikey.com/product-detail/en/murata-electronics-north-america/GCJ188R71H104KA12D/490-5798-1-ND/2783823)
For C46, C89, C144, C145, C146, C147 I used CL21B105KAFNNNG (https://www.digikey.com/product-detail/en/samsung-electro-mechanics/CL21B105KAFNNNG/1276-6469-1-ND/5958097)
For C53 I used CC0805JRNPOABN100 (https://www.digikey.com/product-detail/en/yageo/CC0805JRNPOABN100/311-4245-1-ND/8025334)
So the only one of those substitutes in the pre-amp and PA RF chain is the .1uF caps where used as a coupling/DC blocking device. But in this role it's RF current and Q capabilities should be far less critical critical then if it were in a tuned circuit. I don't think it would be worth swapping those out, do you?
Also, I just finished up the second Hermes Lite for my brother last night. I didn't have enough of the 22AWG PTFE wire on hand so I wound T3 with 24AWG PCV insulated wire instead. The power output is even lower on this board, barely over 3 watts on all bands except around 14mhz. I'm grabbing more 22AWG wire today to rewind T3, hopefully that's the reason. I think I have an extra BN43-202 core laying around here somewhere, I may try switching to that core too.
Cheers!
Dieter - W3DRK
On Friday, June 1, 2018 at 6:20:12 PM UTC-4, Steve Haynal wrote:
Hi Dieter,
in3otd
Jun 5, 2018, 1:16:46 PM6/5/18
to Hermes-Lite
Hello Dieter,
I agree that the different 0.1 uF caps should not have any effect on the circuit performance. Could you measure the low-power output over frequency also for the second H-L? I think the wire diameter for T3 is really not critical.
It may be useful to measure also the output directly at the AD9866 TxDAC on DB3 and check the R69 resistor which sets the TxDAC output current.
I agree that the different 0.1 uF caps should not have any effect on the circuit performance. Could you measure the low-power output over frequency also for the second H-L? I think the wire diameter for T3 is really not critical.
It may be useful to measure also the output directly at the AD9866 TxDAC on DB3 and check the R69 resistor which sets the TxDAC output current.
73 de Claudio, IN3OTD / DK1CG
Dieter Kedrowitsch
Jun 6, 2018, 8:48:03 AM6/6/18
to Hermes-Lite
The problem of extremely low TX power on the 2nd Hermes Lite I built has been solved. The issue was a tiny solder bridge between one of L8's pads and an adjacent via to ground. This grounded one leg of the AD9866's TX output. Picture attached.
With that problem resolved, I was able to compare the low-power TX between the first Hermes Lite 2 I built and this second one. Very similar to the first Hermes Lite, the output is between about 15.5 and 16.5 dBm. And just as before, dropping R55 to 100 ohms bumps this up to over 17dbm which allows the PA to make well over 5 watts on all bands.
Cheers!
73 de W3DRK
With that problem resolved, I was able to compare the low-power TX between the first Hermes Lite 2 I built and this second one. Very similar to the first Hermes Lite, the output is between about 15.5 and 16.5 dBm. And just as before, dropping R55 to 100 ohms bumps this up to over 17dbm which allows the PA to make well over 5 watts on all bands.
Cheers!
73 de W3DRK
Steve Haynal
Jun 8, 2018, 9:13:32 PM6/8/18
to Hermes-Lite
Hi Dieter,
I will try to add measurement data to the wiki. We can then compare what you measure with what I measure at several placed in the TX chain.
73,
Steve
KF7O
Steve Haynal
Jun 8, 2018, 9:35:38 PM6/8/18
to Hermes-Lite
Hi Dieter
This is when we discovered and fixed low 10M output power. There are many good reference measurements in that thread. We should add a summary table to the wiki.
73,
Steve
KF7O
Reply all
Reply to author
Forward
0 new messages