Update on Spurious Emissions Experiments

[Steve Haynal]

Hi List,

Since I am getting some meaningful data from my scope as a spectrum analyzer setup, I've been experimenting to determine the causes of the various spurious emissions. First, there are the expected harmonics of the fundamental. Second, there are combinations of nFs +/- mFtx. I've focused on the second case, and in particular the spur at 2Fs - 5Ftx =  22.981 MHz when Fs is 73.728 MHz and Ftx is 24.895 MHz. Here is a summary of what I've learned:

Natural Harmonics:

• Lowering the power by any means helps with the expected harmonics of the fundamental. We can lower some of these harmonics by an additional 7-10 dBc when lowering the output power.

nFs +/- mFtx spurs, in particular the one at 22.981 MHz:

• When power is lowered by reducing the input audio signal, many of the spurs are cleaned up. It appears that not using the most significant bit to the DAC helps.
• When power is lowered by settings in the AD9866, there is little effect on the nFs +/- mFtx spurs. For example, the PowerSDR drive setting has little effect.
• Adding dither to the data sent to the DAC does not help.
• A synthetic signal generated by the FPGA to eliminate the DSP chain produces the same results. This is a nice confirmation that the DSP chain (cordic/NCO, filters) is pretty clean. 
• There is no meaningful difference between 5V and 3.3V to the center tap of the TX balun. This is good news.
• Various loads on the AD9866 have little effect on these spurs. I have tried 17 Ohms to 88 Ohms in 4 or 5 increments. As the power is reduced, the level of the natural harmonics reduces faster, but the nFs +/ mFtx spurs stay about the same relative to the carrier.
• Various ways of splitting the gain between IAMP and DAC do not effect the spurs much.
• The firmware now supports 79.872 MHz. The spurs move up in frequency but do not diminish much.
• LPF reconstruction filter does not do much for the spurs appearing within the filter pass band.

I still need to experiment with currents via Rset and the standing current register. Also, I want to measure how clean the output is directly from the TXDAC with no IAMP. I measure the 22.981 MHz spur at around 55 dBc. This is not terrible, but I'd like to see it 60 dBc or more. Hopefully this week I will settle on a configuration that minimizes these spurs.

73,

Steve

KF7O

[ZL2APV]

Thanks Steve, great research.

73 Graeme zl2apv

[John Laur]

Harmonics and phase noise from the clock are a perhaps a possible
contributing factor to run down as well.

73, John K5IT

> --
> You received this message because you are subscribed to the Google Groups
> "Hermes-Lite" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to hermes-lite...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[irbsu...@yahoo.co.uk]

Hi Steve,
Very encouraging results, I am looking into getting some 73.728MHz clocks at the moment, should I get ones at 79.872 MHz instead?
Any issues with that?
Thanks,
Andrew
G4XZL

[Steve Haynal]

Hi Andrew,

It depends on how soon you want to upgrade. Eventually the 79.872 MHz will the be the better choice. Right now the following needs to be done:

• The receiver CICs have been rebalanced. This still needs to be carefully tested and amplitudes adjusted.
  
• The amplitudes with the new TX have to be adjusted.
• A full duplex interface runs at almost 160 MHz, which may be too fast for the BeMicro+Hermes-Lite configuration. I've only tested half-duplex which runs at 79.872 MHz.

The above will hopefully be done in the next month or two as part of the v2 development.

73,

Steve

KF7O

[James Ahlstrom]

Hello everyone,

I made some more measurements.  First I looked at Steve's spur at 2Fs - 5Ftx =  22.981 MHz when Fs is 73.728 MHz and Ftx is 24.895 MHz.  I changed Ftx to 24.900 and 24.905 MHz.  The spur moved to 22.956 and 22.931 MHz.  This is five times the change in Ftx as predicted by Steve's equation.

Next I tried changing the Tx level both by changing the TxLevel digital gain control, and then the digital data by means of the Quisk spot level.  As he said, lowering the digital data helps and the TxLevel does not.  Unfortunately, lowering the digital data results in a loss of bits; that is, a 6 dB reduction in data level means we don't use the 12 bits of the AD9866, we only use 11.

Next I tried using the TxDAC alone without the IAMP.  This required two changes, both in rtl/myhdl/ad9966.py.  I un-commented the code to turn on the TxDAC in register 0x0E,

and I removed the code to set the Tx level in register 0x0A.  The Tx level assumes the IAMP, and the code is incorrect for the TxDAC alone.  The loss of this code means the TX level control is disabled, but the default is full power.  I then installed the circuit in Figure 63 of the data sheet, with a 1:1 transformer and Rs not installed (no back termination).

The measured output at 24.895 MHz is 5.6 dBm, and at 29.0 MHz it is 5.5 dBm.  The data sheet claims 10 dBm peak output with a 2:1 Z transformer.  So 10 dBm less 3 dB for peak to RMS, and -3 dB for the 1:1 transformer, and + 1.9 for the increase in current we are using gives 5.9 dBm, close to what we are seeing.

The spur at 22.981 now measures -71 dBm, an excellent result.

The 44.728 MHz spur for Ftx 29 MHz is -44.5 dBm, or -50 dBc.  This is similar to what was measured before.  This spur is probably due to the 2X interpolation filter instead of a distortion mechanism.  We can hope that increasing the clock will drive the spur into the filter stop band and produce a good improvement.  Anyway, this spur is filterable.

The output of the TxDAC looks very clean, but there is a lot of junk at frequencies above 100 MHz.  The signals at 2Fs +/- Ftx are -3.3 and -12.1 dBm.  There is a signal at 148.4 MHz (approx) of -47 dBm with the key up.  There are many signals above 200 MHz that we need to get rid of.

The output of 5 dBm seems large compared to the 10 dBm output of the IAMP, but remember that the TxDAC is not back-terminated, and the IAMP is.  So terminating the TxDAC would cost -6 dB.  But there is no need to terminate the TxDAC if it is buffered by an on-board amp to get to the required 10 dBm.

Jim

N2ADR

[Steve Haynal]

Hi Jim,

Thanks for the data. It is good to know that things are cleaner with no IAMP. I tried lowering the current by increasing the value of Rset. This lowered the power in the same way as changing the TX Level did - improvement with natural harmonics, but no improvement with the 22.981 spur. I still want to cut this current in half and then double the standing current of the IAMP. I hope that will have the same effect as not running the DAC near fullscale, but still provide 12 bits of resolution and around 14 dBm out.

73,

Steve

KF7O

[James Ahlstrom]

Hello Group,

Here is a photo (sorry, old instrument) of the spectrum of the TxDAC output (not the IAMP) from zero to 500 MHz.  The top line is 10 dBm.  The marker is the 29 MHz output.  The screen shows 34 MHz, but frequency readout is not too accurate with my Anritsu MS2670A.  There are strong outputs up to and past 500 MHz.  Perhaps that is the reason the IAMP is producing strong spurs.  There should be a low pass filter at the TxDAC output to protect the IAMP, but there is no way to do that with the AD9866.

Jim

N2ADR

[ZL2APV]

Hello Steve, Jim and Group,

Very interesting work you are doing and your results are changing the way I am thinking now. Currently the 30 MHz roofing filter on John's amplifier board was tweaked to get rid of the spur that occurred at 34? Mhz. With a shift to a 79.872 MHz oscillator or even an Si570 variable oscillator I think the need for this filter may be gone and perhaps it would be better to redesign it as a 7 pole Cauer for steep cutoff, adequate attenuation out of passband and excellent S11 for good in band impedance characteristics. It won't require any changes to the PCB or toroids but of course may need some smd caps to be altered. Would appreciate any comments.

73, Graeme zl2apv

[in3otd]

Hello Jim,
thanks for the interesting measurements; if the TxDAC output is really a current source as shown in the datasheet diagram it may be possible to apply some filtering with a network between IOUTP+/IOUTP-, a capacitor in parallel or something more complex. The IAMP input impedance might be relatively low but I guess not too much.

73 de Claudio, IN3OTD / DK1CG

[James Ahlstrom]

Hello Graeme,

I don't think the 5-pole Chebyshev filter with the trap at 44 MHz is appropriate even for the 73.728 oscillator because it has insufficient attenuation at 118.46 MHz.  I am using a 7-pole Chebyshev design and today I will finish up and publish my measurements.

Jim

N2ADR

[James Ahlstrom]

Hi Claudio,

I hadn't thought of that.  Interesting, but only one pole of filtering is available.  My current thinking is that the TxDAC outputs should be used instead of the IAMP, there should be a balanced 5-pole Chebyshev filter directly at the TxDAC output, and there should be a subsequent buffer on the board.  But that will take a while, and meanwhile we need to get all those 5W amps going.  I will publish more measurements today.

Jim

N2ADR

[ZL2APV]

Yes that was my concern also which prompted me to consider a 7 pole Cauer. I will wait for your design as there is no need for me to proceed if you are already doing it.

Cheers, graeme

[James Ahlstrom]

Hello Graeme,

I may have misunderstood your question.  I was referring to the reconstruction filter on the new front end boards.  You may be referring to the Cauer Tx filters.  Both are required.  I am getting a little confused here.  I will post more measurements today to add^H^H^H lessen the confusion.

Jim

[Steve Haynal]

Hello Group,

After a few more experiments, I've come to some conclusions at least for the 22.981 MHz spur (and other similar spurs) when transmitting at 24.895 MHz. I doubled the standing current for IAMP by changing an internal AD9866 setting but saw perhaps minimal (1-2 dB) improvement of the 22.981 spur. Since the biggest difference in this spur has happened only when changing the range of values sent to the DAC, I believe this and some other spurs are due to non-linearity of the DAC as it approaches maximum range (compression). Essentially, a change in value from 2000 to 2001 (near the maximum) results in a smaller output voltage change than the same step from 200 to 201. This non-linearity is known to cause harmonics that occur all the way up in higher Nyquist zone. These harmonics are then aliased or folded back into the first Nyquist zone where we see them. There are many references which describe this in better detail, including pages 5 and 6 of this one. From Jim's experiments with no IAMP (which I still want to duplicate), this DAC non-linearity appears to be aggravated by the IAMP. Some of the other spurs that Jim recently noted may be other aliases.

Fortunately, in all my measurements this spur at 22.981 MHz was quite consistent, even when turning many knobs. To me this implies that it is a good candidate for predistortion to "undo" the distortion added by the DAC. This may be as simple as adding some emphasis to the signal as it approaches the extremes of the DAC range. For example, scaling by something more than 1.0 as we approach the extremes. It has to be simple to implement easily and nicely in the RTL. This is one direction I will pursue. See this article for more ideas.

I also experimented with no interpolation. The AD9866 only supports half-duplex with no interpolation. I did not see any benefit from no interpolation. Since Fs is now 73.728 MHz, there is more opportunity for aliases to fold back into a smaller first Nyquist zone. I saw that and they were worse than what we see with interpolation. The spur at Fs-Ftx is much stronger as expected, which makes filtering more difficult.

With interpolation, there is still a spur at 0.5Fs - Ftx which I can't explain yet. For example, this is the ~44 MHz spur many people have noted when transmitting on 10M. Relative to the carrier, this spur does not diminish for any set of knob changes, even when reducing the amplitude range sent to the DAC or providing the DAC with the cleanest synthetic signal. Since it depends on 0.5Fs (73.728 MHz for 2X interpolation), I suspect it is an artifact of the digital interpolation LPF. In my measurements, it is in the 55 to 60 dBc range, and will move higher in frequency and become easier to filter with the new 79.872 MHz oscillator. I have a 79.872 MHz setup and have done some experiments but will post more later.

There are still the natural harmonics of the IAMP. These do improve with lower power from the IAMP. This is typical of any amplifier.

I am also seeing spurs at +64KHz and -64KHz from the carrier on high TX frequencies in 12M and 10M. These are at about -55 dBc. I'm not sure if these are real or an artifact of my hacked test setup. Can someone else confirm? If they are real, they may be due to some modulation of the clock.

It is taking much longer than I anticipated to finalize new frontend ideas, and there is still more to do. One thing that I enjoy about this project is that unlike my day job, there are no deadlines and I can spend time on what interests me the most. So, sorry to those anxious for version 2.0, but I intend to enjoy this project. :) Also, my day job has been *very busy* the past few months and will stay that way for at least another few months. Here is what I plan to focus on in the near term:

• Add predistortion emphasis to see if spurs can be reduced.
• Experiment with no IAMP. A weakness of the AD9866 is that it is impossible to put any good filter between the DAC and IAMP.
• Update the frontend card to use a THS6226 connected directly to the TxDAC with no IAMP.
• Cleanup, test and finalize 79.872 MHz oscillator version.
• Experiment with and confirm other's results for a reconstruction filter.

73,

Steve

KF7O

  

 

  

[Alan Hopper]

Steve,

I'd been thinking about predistortion for this, as a simple start I planned to try a lookup table just before the dac output and use the raw adc bandscope data for feedback combined with returned dac values (maybe interleaved with the bandscope data). This does not address phase error but would be an interesting first step.  Just playing with the lut with no feedback might be enough to tell us if it is worthwhile.  I afraid time and lack of fpga skills have stopped me trying this so far.

The THS6226 sounds very interesting and would result in a very neat v2 if it lives up to expectations.

73 Alan M6NNB 

On Saturday, October 31, 2015 at 11:36:41 PM UTC, Steve Haynal wrote

[James Ahlstrom]

Hello Steve,

I tried to find your +/- 64 kHz sidebands by testing at 24.895 and 29.000 MHz output frequencies.  My Anritsu spectrum analyzer shows nothing within a few 100 kHz of the Tx signal.  My noise floor is about - 80 dBc, so I think these spurs are not real.  Someone else please confirm.

When you get that far, I would like to know the amplitude of the Fs - Ftx spur with the 79.872 MHz clock.  I am hoping it will not just move up in frequency, but will also weaken because it will move further into the interpolation filter stop band.  It would be easier to build a 5-pole reconstruction filter because the two inductors are equal.  I also have some other ideas on the reconstruction filter once we get that far.

Jim

N2ADR

[Stew KF5KOG]

I am seeing the spurs at ~+-55khz.  For me 10M is the worst.  They decrease by a couple db as you decrease frequency.  Level difference between drive slider at 41% and 100% is again a couple of db, with 41% being worse case.

My setup is using the Pre-V2 Amp running into a dummy load and -43db tap.

HL drive level is approximate into the Amplifier.

73, Stew

KF5KOG 

[Steve Haynal]

Hi Stew and Jim,

Thanks for the measurements. Here are all the differences I could think of which might explain the different results:

• Stew is using the pre-V2 amp. Do you still see the spurs when barefoot?
• Stew's results at +/- 55KHz are suspiciously different from mine at +/- 64KHz to make me think Stew is using a 61.44 MHz oscillator. But I think I've asked before and Stew is using the 73.728 MHz oscillator. Is that correct?
• Jim is using the surface mount 1:1 transformer, while I and I believe Stew is using a wind your own similar to the original specification of the Hermes-Lite.
• Jim's spectrum analyzer is a more traditional model, while Stew and I are using sampled with FFT approaches. Jim, are you certain your setup can distinguish signals with 64 kHz separation?
• I am using the power supply PCB from Dave Miller. What power supplies are you using Jim and Stew?

Let me know if there are more differences. I will experiment with these differences to see what might explain these spurs.

73,

Steve

KF7O

[Steve Haynal]

Hi Alan,

I had the same thought in the shower yesterday. Essentially we can build a map from "true" dac output values to "distorted" values such that the voltages out of the DAC and/or amp is linear. I am considering putting such a map in the RTL as the RAM requirements are manageable. I think I can do so in such a way that there are still no unused codes. This may also be possible on the software side. Although it is static and simple, it would be interesting to see how far it gets you. 

As part of my recent testing, I have things setup to generate a synthetic sine wave at ~24.990 MHz which is very clean. I have scripts which generate the RTL for that. As a first easy step, I will try to distort this sine wave based on the FFT I am seeing and see if there is any good effect.

73,

Steve

KF7O

[Steve Haynal]

Hi Jim,

With the 79.872 MHz clock, and Ftx of 24.895, I see the Fs-Ftx spur at 54.977 MHz and measure it between 49 and 50 dBc. Note that 54.977 MHz is higher than the second harmonic at 49.79 MHz. At high power levels, that harmonic can be stronger. Also, with Ftx=24.895, I still see the 2*Fs-5*Ftx but now at 35.269 MHz. This can be as bad as 53 dBc if the full range of the DAC is used. 

When Ftx=29.690, I see the Fs-Ftx spur at 50.182 MHz and measure it again at around 50 dBc. I do not see any problematic 2*Fs-5*FTx, probably because it lies at a different place in the sinc envelope of the DAC.

73,

Steve

KF7O

[Glenn P]

James, and all,

I can confirm that... (not real)

I just did the test at both frequencies using the SA here. ie no +-55KHz spurs (or +-64KHz)

Running PowerSDR
Linear power supply to setup.
30MHz reconstruction filter following Hermes Lite.  (ie no amplifiers.)

The attached are all at 50% drive but no changes noticed when running 100% drive either.

Pics of 24.895 and 29MHz. Also one pic is zoomed right in.   300 Hz resolution on SA so most things should be visible, if they are there !

glenn
vk3pe

[Glenn P]

setup has 30dB wide band attenuator out of H-L also........

glenn

[James Ahlstrom]

Hi Steve,

Actually I am using a 1:2 Z plug-in transformer, a 100 ohm back-termination, and a 7-pole filter on a new front end board.  Attached is a picture.  Yes, this is a non-digital spectrum analyzer.

My power supply is a quality bench supply at 8.1 volts feeding 5 and 3.3 volt linear regulators.

I think I could see any sidebands.  To make sure, I changed the span to 5000 Hz, and looked at the Quisk IMD output.  I saw two peaks.  Then I searched by hand around 55 and 64 kHz and saw no output.

Jim

N2ADR

[James Ahlstrom]

Hi Steve,

On Monday, November 2, 2015 at 4:59:14 AM UTC, Steve Haynal wrote:

When Ftx=29.690, I see the Fs-Ftx spur at 50.182 MHz and measure it again at around 50 dBc.

 

That is disappointing.  I measured 49.4 dBc with the 73.728 MHz clock, so raising the clock did not help the interpolation filter as I expected.

Still, the increase in frequency from 44 to 50 MHz is helpful because both the reconstruction filter and the transmit filters will help eliminate this spur.  We can focus on a 5-pole Chebyshev filter for version 2.  Obtaining the parts for a 7-pole is much harder, and should not be necessary.  I still think cheap 0805 inductors are fine for this filter.

By the way, I measured the response of the 7-pole Chebyshev by itself without the transformer or AD9866.  The response from 0 to 500 MHz showed many peaks and valleys, and wasn't anything like the predicted response above 150 MHz.  The next board needs a solid uninterrupted ground plane across the filter and UHF construction techniques.

Jim

N2ADR

[Stew KF5KOG]

It looks like I may be picking up those signals from somewhere in my setup/equipment.  I will see if I can find the cause.

I am using the 73.728 oscillator.

73, Stew

KF5KOG

[in3otd]

Jim,
for my understanding, the filter is built on the V2 frontend (C1-C4, L1-L3) board, right? Looking at the layout of that part, the grounding seems not to bad to me, but I may be underestimating what is actually needed for a good rejection at UHF. Might some of the filter peaks there be due to the inductors resonances? BTW, which inductors are you using, the AIML-1206 series?

73 de Claudio, DK1CG

On Monday, November 2, 2015 at 4:34:42 PM UTC+1, James Ahlstrom wrote:
[snip]

[James Ahlstrom]

Hello Claudio,

Yes, I am building on the latest front end board.  The 330n inductors are Bourns CW252016-R33J, Q=45, package 1008 (not 0806 as I said previously) self resonant 570 MHz.  The 360n inductor is TE Connectivity 36502AR36JTDG, Q=48, package 0805, self resonant 580 MHz.  Yes, higher peaks can be inductor resonances.  But I suspect the board plays a role.  There should be a solid wide trace from the transformer ground to the output ground with all capacitor grounds attached to it.

For the final version, the filter should go before the transformer, not after it.  The transformer will not be flat to over 500 MHz, and its reactance will reflect power back into the AD9866.  Since the filter goes before the transformer, we need a balanced design.  A balanced 5-pole Chebyshev with four inductors of small size should be easy.  All inductors have the same value.  We just need to pick a cheap inductor that is easy to find.

Jim

N2ADR

[in3otd]

Thanks Jim,
since the inductors have a quite high resonant frequency, they are likely not the (main) cause of the filter peaks above 150 Mhz, so it could be the PCB...
By the way, I have played a bit with an optimizer and saw that you could build a very similar filter using 100p, 330n, 200p, 390n, 200p, 330n, 100p, which avoids the not-so-common 360 nH inductor.

Regarding the filter positioning, before or after the transformer, will it really make a difference in practice? I understand that currently, at high frequency, the DAC output "sees" mainly a high impedance from the leakage inductance of the transformer, while having the filter first it will see mainly the low impedance of the first shunt capacitor, but I do not remember having seen a differential filter before the transformer in the manufacturers documents for DDS or ADCs when a single ended output was needed, so maybe in practice the high-frequency loading has little effect. Will be nice to do some measurements, hi.

73 de Claudio, DK1CG

[ZL2APV]

That is a nice filter with easy values. RL better than 18 dB and Q is really only affecting insertion loss. Running it as a balanced floating filter means 6 inductors but the 50 and 100 pF capacitors are still good standard sizes. The response of the filter hardly changes with terminations ranging from 10 to 1K ohms.

73, Graeme zl2apv

[Steve Haynal]

Hi Stew,

What I saw are artifacts from my scope/spectrum analyzer setup. Changing settings on the scope effects these spurs. I don't see them on Quisk. 

73,

Steve

KF7O

[Steve Haynal]

Hi Jim,

I went with a 2 layer board for this frontend card but this will eventually be on the 4 layer main board and include a better ground under the filter.

All the capacitors are on the back. You could run a copper braid or wire from the transformer ground, over and soldered to the capacitors and then to the output ground to see if this improves things.

73,

Steve

KF7O

[Steve Haynal]

Hi Claudio,

Another reason for the filter positioning is that a balanced filter would work well with an op amp stage.

73,

Steve

KF7O

[Glenn P]

I knocked this one up for a quick test.....  used 0603 SMD inductors and 0603 caps.  The 200pF actually 180 + 22pF. Otherwise as per Claudio's design.

The 0603 inductors have a Q of 25.

The peak on the right is due to coupling between coax cables in the test setup. I can reduce the peak by moving the cable around.

glenn
vk3pe

[John Williams]

Glenn, et.al.

I built the TX reconstruction filter using the AIML-1206 inductors. The Q of these are listed at min of 35.  I did not have any 200pf caps and did not want to stack 100's (lazy I guess). The larger inductors fit very nice.

On my v2 frontend board, I did the following -

1. I plan to not use the T/R relays U1 and U5. I wired it straight from T1 to L1. T/R is handled by my amp.
   
2. I wired the TX reconstruction filter as 100p, 330uh,220p,390uH,220p,330uH,100p using AIML-1206 inductors. Fit nice.
   
3. I plan to eventually populate U2,U3 and U4 but for now I have bypassed them, feeding T2 to P3. RX filter not populated.
   
4. P2 will be used for a power tap from the V2 amp board, but now left unpopulated.
5. I left P5 and P6 unpopulated
6. I have a header for CN3 but do not plan to connect it until Steve has his logic implemented for it.
   
7. I put a 2x6 header in CN2 and will use a 12 conductor ribbon cable to connect to the same style connector on the V2 amp board. The pinouts are compatible.
8. I used BN-43-2402 hand wound transformers for T1 and T2. T1 was wound according to the prior frontend. T2 was wound per Steve's note with T2 (8:1 with no centertap, 14 turns on PCI-E side, 5 turns on antenna side). Jim Ahlstrom has been using some commercial transformers. See his notes. I listed some that Steve suggested in the BOM attached.
   
9. I used 2 pin molex headers for P1 and P3.
10. I placed a leaded 100 ohm resistor across the input side of T1, per the suggestion of Jim Ahlstrom. This will limit the outout of the frontend to around +10dbm max.
    

For the 3 new builders of the pre-v2 amp, I will publish a new BOM tomorrow that will have updated part numbers for the filters and the regulator bias changes.

There are no build instructions. You will have to follow the schematic to build. Follow the V1.0 schematic for everything except the filters. For those you will follow the V1.1 schematic. At this point, we are not firm on the 17/15 filter values. We may do away with bank6 if we get a 17/15 filter that we think is easy to reproduce. You may need to order some T37-10 toroids, maybe 6-9 of them. They are pretty inexpensive. We may be using them on the 17/15 and 12/10 roofing filter.

You will need to make 2 changes.

One is to create a power tap. I will send out a pic of that. Easy mod to P3 where you cut the trace from K1 pin 1 that goes to P3 pin 1. Then run a short jumper from P3 Pin 1 to the ground side of C14 that is right nest to P3.

Second one is to make a change to allow the bias voltage to increase to 6V. Look for Glenn's note on Bias change.

There are 3 options for T1. Order 2 BN-43-202 binoculars. The first option is described in the wiki for the superband build. The second option uses 2 binocs stacked with the coax that was tested by Stew and noted more consistent output power. Look for his note. Jacinto also had a pic of the implementation recently. Third is one using turns of enameled wire proposed by Claudio. I am going to try that one.

It is probably time for me to make a small errata document with pics of all of the mods we have made so far. Will work on that also.

John

[ZL2APV]

Hi John,

Great information and will be very helpful when my bits arrive. Just to clarify a point, you said "Follow the V1.0 schematic for everything except the filters. For those you will follow the V1.1 schematic." The V1.0 schematic does not have the bias mod but I guess you mean for everyone to implement that so perhaps you would like to change the instructions to say "everything except the filters and bias mods".

73, Graeme zl2apv

[John Williams]

Yes, Graeme, good point.

--
You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[James Ahlstrom]

Hello Claudio,

I think putting a balanced filter before the transformer may be better because the AD9866 output is a current source, so the initial capacitor limits the voltage generated at high frequencies.  And thanks for the new filter design.

Jim

[John Williams]

We now have a wiki article on this topic

[Glenn P]

John

finding a 2x6 ribbon cable seems to be elusive here.  10 & 14 pins are much more common. Is it possible to look closer at this availability, please

glenn

[Glenn P]

I meant the actual female connectors of course, not cable

glenn

[Sid Boyce]

I have these Ebay items on order from China as I couldn't find them
locally in UK.
Item Number 181021915592
Amazon.com has them but won't ship to UK.
73 ... Sid.

On 05/11/15 21:32, Glenn P wrote:
> I meant the actual female connectors of course, not cable
>
> glenn
>
>
> On Friday, November 6, 2015 at 8:31:40 AM UTC+11, Glenn P wrote:
>
> John
>

> finding a 2x6 ribbon cable seems to be elusive here. 10 & 14

> pins are much more common. Is it possible to look closer at this
> availability, please
>
> glenn
>
> On Thursday, November 5, 2015 at 11:56:26 AM UTC+11, John Williams
> wrote:
>

> Glenn, et.al <http://et.al>.

>
> I built the TX reconstruction filter using the AIML-1206

> inductors. The Q of these are listed at min of 35. I did not

> have any 200pf caps and did not want to stack 100's (lazy I
> guess). The larger inductors fit very nice.
>
> On my v2 frontend board, I did the following -
>

> 1. I plan to not use the T/R relays U1 and U5. I wired it

> straight from T1 to L1. T/R is handled by my amp.

> 2. I wired the TX reconstruction filter as 100p,

> 330uh,220p,390uH,220p,330uH,100p using AIML-1206
> inductors. Fit nice.

> 3. I plan to eventually populate U2,U3 and U4 but for now I

> have bypassed them, feeding T2 to P3. RX filter not populated.

> 4. P2 will be used for a power tap from the V2 amp board, but
> now left unpopulated.
> 5. I left P5 and P6 unpopulated
> 6. I have a header for CN3 but do not plan to connect it

> until Steve has his logic implemented for it.

> 7. I put a 2x6 header in CN2 and will use a 12 conductor

> ribbon cable to connect to the same style connector on the
> V2 amp board. The pinouts are compatible.

> 8. I used BN-43-2402 hand wound transformers for T1 and T2.

> T1 was wound according to the prior frontend. T2 was wound
> per Steve's note with T2 (8:1 with no centertap, 14 turns
> on PCI-E side, 5 turns on antenna side). Jim Ahlstrom has
> been using some commercial transformers. See his notes. I
> listed some that Steve suggested in the BOM attached.

> 9. I used 2 pin molex headers for P1 and P3.
> 10. I placed a leaded 100 ohm resistor across the input side

>> I knocked this one up for a quick test..... used 0603 SMD
>> inductors and 0603 caps. The 200pF actually 180 + 22pF.

>> Otherwise as per Claudio's design.
>>
>> The 0603 inductors have a Q of 25.
>

> --
> You received this message because you are subscribed to the Google
> Groups "Hermes-Lite" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to hermes-lite...@googlegroups.com

> <mailto:hermes-lite...@googlegroups.com>.

> For more options, visit https://groups.google.com/d/optout.

--
Sid Boyce ... Hamradio License G3VBV, Licensed Private Pilot
Emeritus IBM/Amdahl Mainframes and Sun/Fujitsu Servers Tech Support
Senior Staff Specialist, Cricket Coach
Microsoft Windows Free Zone - Linux used for all Computing Tasks

[Glenn P]

thanks Sid
I ordered some also to avoid the problem on Pre V2 at least.

glenn

[John Williams]

Same store has the complete cable in 2 lengths...

171420139585

John

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

To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.

[Sid Boyce]

Thanks John,
I was thinking of cables shorter than the standard 20cm, my be 10cm, but
in this application the length is not critical.
73 ... Sid.

>> <mailto:hermes-lite...@googlegroups.com>.

>> For more options, visit https://groups.google.com/d/optout.
>

> --
> You received this message because you are subscribed to the Google
> Groups "Hermes-Lite" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to hermes-lite...@googlegroups.com

> <mailto:hermes-lite...@googlegroups.com>.

> For more options, visit https://groups.google.com/d/optout.