|
Encouraged by the results of the [80m/60m] filter I implemented my [40m/30m] filter to go with it. The attached VNA sweeps include the LPF both diode switches and the coupler. I was about to measure Tx harmonics when I dropped a crocodile clip on the back of the PCB. I guess there's a law about these things but it landed across the 12V DC input jack and the 5V Vdd pin. Entirely my fault but fireworks and a fried QMX as would be expected. So no more work on this, but if anybody wants details of either LPF just let me know.
|
Are you redesigning all LPFs to squeeze out a room for a high band filter?
Did you measure the stock 60/40m and 30/20m LPF responses, or is it generally published? I'm planning to use those as-is and design 17/15/12/10m LPF (I guess 11m could be counted to inflate the catalog spec but I have no plan of being there).
|
When I first bought the kit I intended to redesign it with LPFs for [40m/30m], [20m/17m] and [15m/12m/10m]. But when I realised that [80m/60m] was also possible I decided to do that instead - along with [40m/30m] and [20m/17m/15m]. So yes I was trying to make more bands. At no time did I ever build with the stock LPFs.
|
Ok- we have a bit different approach to similar goals. My thought: when I get on 80m I will most likely have a bit more serious equipment anyway, so I would gladly give up on 80 in exchange for 17-10m. Plus, if I have a 60/40m filter, I could always build an external LPF/BPF for the feedline if I wanted an ultraportable 80m rig.
I'll share my filters when I measure the response.
|
My QMX is now fried. So for me it's 'Game over - no lives left'. I suspect you might have a hard time doing 17m-10m with a single LPF because the second harmonic of 2x18.068MHz isn't many octaves above 29.7MHz. If you're looking for 20dB or so of second harmonic attenuation then groups like [80m/60m], [40m/30m] and [20m/17m/15m] work quite well. As does [15m/12/10m]. It would be nice if there were some kind of 'balance' adjustment on the PA to null out the even harmonics.
|
My design candidate #1 can get -28dB at 36MHz and min ATT above 42MHz is 46dB (fc=29.7MHz).
My candidate #2 is fc=32MHz and -18dB at 36MHz and min ATT is 48 dB. I have a version with fc=26MHz and get min ATT of 60dB if I want to overkill the problem (which is required in some country like JA).
I tested robustness against component value variations but let's see how the input impedance variation shows up in the actual emission spectrum.
|
Hi Nick,
I¡¯m interested in your filter design. I¡¯d like to Spice it for comparison. Do you have a schematic available?
Thanks,
Tony - AC9QY
Encouraged by the results of the [80m/60m] filter I implemented my [40m/30m] filter to go with it. The attached VNA sweeps include the LPF both diode switches and the coupler. I was about to measure Tx harmonics when I dropped a crocodile clip on the back of the PCB. I guess there's a law about these things but it landed across the 12V DC input jack and the 5V Vdd pin. Entirely my fault but fireworks and a fried QMX as would be expected. So no more work on this, but if anybody wants details of either LPF just let me know.
|
Hi Tony,
For the [40m/30m] filter the basic ladder components running from the PA towards the coupler are:
470pF, T30-6 15 turns, 680pF, T30-6 13 turns, 330pF.
The capacitor in parallel with the second toroid is 150pF.
According to my measurements the toroids would seem to be 830nH and 647nH respectively, although for my own simulations I used the s1p files that I extracted using 'nanoVNA-saver'.
- Nick.
|
Sorry, the second inductor should have been 12 turns (620nH).
|
|
|
There's scope to improve the response by scaling the values to shift the corner frequency a little lower. That would improve the harmonic attenuation whilst bringing the regions of lowest loss and best return loss on to the wanted 40m and 30m bands.
|
Is there a need to improve to meet -43dBc max harmonic or the UK emission standard? I guess the answer depends on how balanced the push-pull is?
JA requires -60dBc but they may not give a license to a transmitter that can transmit off-band including 60m (not allocated to amateur service in JA), so designing a filter bank for them is only a thought exercise at this point.
|
Ryuji,
I was about to start measuring harmonics when I toasted my radio so I cannot speak about existing performance. In redesigning the filters I was simply hoping that that the original design was OK, and that my new filters were no worse than the originals.
Performance requirements of amateur equipment in the UK are stated in Ofcom document 'UK Interface Requirement 2028'. But that itself doesn't contain much technical detail. It simply cites European standards such as EN 301 783 which in turn specify the limits. See link below.
Although my board is now toast I might still be able to work on the LPFs, in which case I might do some more on this.
- Nick.
|
Tony,
If it helps with your SPICEing I've attached a screenshot of the complete circuit definition that I used in my QUCS simulation. It includes modest allowance for strays and losses. My 0.5ohm for the diodes is just a guess. Note that here the .s1p blocks are de-activated and the toroids are simply specified as lumped inductors.
The four 'opt' blocks on the far right show component values results derived from some 'quick and dirty' optimisations, whereby I optimise the 14MHz rejection for a number of different passband return losses. The first number in the title of each block? is the notional target return loss (in dB) and the second is the 14MHz rejection achieved. (These results are not rigorous because the R.L. is only assessed at a small number of frequencies. But it gives you an idea of the possible trade-offs.)
The issue of acceptable LPF return loss might be important given the number of PA failures reported under poor VSWR conditions. Because depending on the antenna load phase the LPF's own return loss function (whose own spec assumes a perfect 50ohm load) may worsen the VSWR seen by the PA.
- Nick.
|
Nick, that makes sense and I'm sorry to hear what happened on your board.
Your measured S21 looked good enough for the US standard assuming no harmonic is stronger than the fundamental and the 40m 2nd harmonic is weak enough. I guess the latter is the unknown here. (and there is always a small chance of one of the push-pull transistor blows without the operator not noticing for some time)
For my high band filter, I use elliptic function filter to achieve steeper transition band without modifying the board. I can simply add a 2012/0805 SMD cap on the bottom side. But I'll first build the filter on barrack and measure with NanoVNA V2 before I build the kit though.
|
Ryuji,
The response and values that I posted to Tony are OK but better filters are possible. For [40m/30m] this is better:
390pF, T30-6 16 turns, 620pF, (T30-6 13 turns//150pF), 330pF.
The attached simulation sweeps show 17dB R.L> and 24dB second harmonic rejection.
This still isn't quite right but it's closer. The real problem isn't in deriving the ideal values for the best filter, but in going from those to a filter that can be made with the available inductor and capacitor values.
|
?
?
Nick and Ryuji,
Here¡¯s my SOTA 30/40 LPF design it measures quite well after building
73
Ross
Sent from my iPhone 6
|
Hi Ross
I'm guessing you've done as much work on this topic as anybody. Do you have any idea of the inherent even-harmonic rejection of the QMX PA. If we knew that then we'd have a better idea of what filtering is actually required. I'd measure my own but my entire QMX is now toast.
- Nick.
|
Nick,
I only have a TinySA to test my LPFs with and the rejection but I did get some good plots from this. Usually, Hand builds my designs after send him the info and he checks on his up-market spectrometer. He hasn¡¯t tested this yet.? If I can find you my TinySA screenshot amongst the plethora of pics I¡¯ll post.?
Mind you, the LPF was tested on a 12V RWTST in the QMX.?
73
Ross
toggle quoted message
Show quoted text
On 21 Sep 2023, at 09:43, Nick G4IKZ via groups.io <nds12321@...> wrote:
?Hi Ross
I'm guessing you've done as much work on this topic as anybody. Do you have any idea of the inherent even-harmonic rejection of the QMX PA. If we knew that then we'd have a better idea of what filtering is actually required. I'd measure my own but my entire QMX is now toast.
- Nick.
|
Yeah exactly, Nick, if we knew 15dB down at second harmonic is adequate, I could have nicer passband like Ross's for 17-10m LPF. My attitude here is that I can over-build now and slack later. Meanwhile if I get on the air with it I won't feel bad about it.
|
Ryuji Suzuki AB1WX