V1.4 Frontend

[James Ahlstrom]

Hello Group,

I am back from my ski trip.  I am working on the 1.4 frontend board.  It is basically similar to 1.3, but uses a dual op amp similar to the original Hermes.  I will describe the details when I get further along.

Jim

N2ADR

[James Ahlstrom]

Hello Group,

I managed to fit the dual op amp on the same board size.  I had to change the dual 0805/1206 inductor pads to just 0805.  If you can't get the 0805 parts try to get 0603 instead.  I am using the OPA2674 in 8-SOIC.  I looked at the ADA4311 that Steve suggested, but it is only 3 millimeters square, and has a central pad that must be soldered to the board.  I have a hot air machine, but I was worried about whether I could solder it down without a stencil.  I will also try a THS3092. If the design works out, maybe we can try the ADA4311 later.

I see the original Hermes used two stages of op amp.  The issue will probably be how much gain we can get while preserving the large signal bandwidth.  It helps that the original Hermes needed bandwidth to 54 MHz while we only need 30 MHz.

Jim

N2ADR

[James Ahlstrom]

Hello Group,

I ordered my 1.4 boards from Osh Park; not sure when they will get here.

Meanwhile I decided to follow up on a suggestion from Steve.  He pointed out that the dual op amp could drive a following push pull stage for more output.  So I built a class A push pull amp on a scrap of PC board.  It uses two 2N5109 metal can transistors, is biased at 50 ma and 12 volts, and uses a 2:1 turns ratio center taped output transformer.  I am driving it from an input transformer as a test; the final design would be driven from a balanced source.  I intended this as a power stage, but when driven with 740 mVpp at each base, it produces 7.0 Vpp at the output, or 21 dBm.  The 740 mVpp is the same as the output from the TxDAC, so this amp could be used as a replacement for the op amp.  A redesign for more gain and lower bias would be indicated.

Using a discrete transistor amp instead of an op amp was discussed by Claudio.  He build several amps to test this idea.  Mine differs in being a balanced design instead of single ended, and having the transformer at the end instead of the beginning.

The trade off between op amp versus discrete will be distortion.  I think a discrete stage will be fine for transceiver use, as its harmonics will be irrelevant since subsequent amp stages will add harmonics anyway.  But for instrument use, for example, a signal generator, the op amp will produce much lower levels of harmonics.  And the op amp output impedance can be an accurate 50 ohms.

I still want to see how much output we can get from a cheap $4 op amp.  But we can think about some design decisions.  We could use a discrete amp on the basis that an op amp costs more and is hard to obtain outside the USA, and accept a reduction in the instrument utility of Hermes Lite.  Or we could keep the op amp, and add a two transistor output stage as a build option.  Or we could use the op amp alone and require extra amplification to be off board.

What do you think?

Jim

N2ADR

[John Williams]

How much drive will you get from the differential opamp? I have a breadboard push pull amp based on the rd16hhf1 that would be fun to try.

--
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.

[ZL2APV]

In my opinion we should pursue the op-amp. I believe that the output should be a "clean as possible" 50 ohm instrument grade signal if we can achieve it. We should target the level from the 1.4 board to be as high as the op-amp will allow consistent with being clean and use this to set the base level out of the board. The place to get the amplification is on the amplifier board and if it requires a pre-driver then so be it. I suspect there is enough gain available from John's board to still give 5 watts out although it may require the first amplifier to go from 9 dB to 11 dB gain by simply altering the turns of the transformer.

73, Graeme ZL2APV

[James Ahlstrom]

On Wednesday, March 16, 2016 at 5:27:58 PM UTC-4, John Williams wrote:

How much drive will you get from the differential opamp? 

This must be measured, since I don't know how to calculate it.  Asking for too much gain or too much output decreases the bandwidth.  My 1.40 board will accommodate any 8-SOIC dual op amp, and I have two on hand.

Jim

N2ADR

[ZL2APV]

Hi Jim,

When you were off skiing I attempted to make a design spreadsheet for the amp to speed up the design on your return but it turned out to be more difficult than I expected and I eventually abandoned it. I came to the conclusion that it would be far faster to tweak values and measure the results and obviously seems to be the track you are taking and I now understand why :-)

Again I suggest you get a clean 50 ohm output with sufficient bandwidth and measure it. We establish this level as the base level and design following stages from there. In the final final amplifier board design I would like to see push pull stages right through. For an instrument it may be necessary to switch in a supplemental instrument amplifier.

5 watts out from the Hermes-Lite is a big attraction due to a big following of QRP operators and should be catered for. It does not matter what the gain distribution for the unit is so your 1.4 board can set a base level and we get the 5 watts from there. Having prattled on about 5 watts, I believe that we should really go for 10 - 15 watts as this is really necessary to drive a set of transistors to 100 watts which is a level a lot of people will be looking for. The amplifier can have a toggle switched attenuator at the input or other stage to limit it to 5 Watts.

73, Graeme ZL2APV

[Steve Haynal]

Hi Jim,

Thanks for all your work on this. We really are fortunate to have you putting this much effort into the frontend. Your 1.4 board and push/pull experiments really widen the concrete options we have to pick from for V2.0.

I have your 1.3 board and a couple of op amps I want to try, but haven't built it yet.

My plan is to push ahead with the rest of the V2.0 schematic and layout and leave as much space as possible for the RX/TX frontend. When the final area is known, we can pick from the options you and others are testing.

73,

Steve

KF7O

[James Ahlstrom]

Hi Steve,

I am very happy to help.  It is pleasant to work with such a talented group of people.

I am currently working on replacing the voltage regulator with a voltage limiter circuit.  It passes most of the input voltage through up to a limit of 12 volts.

Jim

N2ADR

[James Ahlstrom]

Hello Group,

I finally got my 1.40 boards back and I built the dual op amp buffer.  Initial results look excellent.  The output is 19.8 dBm at 7.2 MHz and 17.5 dBm at 29.7 MHz.  This is with the 7-pole filter and the back termination.  I tested the op amp with my tracking generator, and the -3 dB point is 64 MHz, so there is room for increasing the gain and output.  I measured the output from the AD9866 alone, and it drops only 0.6 dB from 7.2 to 29.7 MHz, so the extra drop must be in the filter or the output transformer.

Next I want to make new 1.41 boards for release.  These will have a center pad for the OPA2677 op amp for greater power.  And I am replacing the voltage regulator with a zener diode and emitter follower since the regulator is hard to source overseas.  I will post more results once I order the new boards and make more measurements.

Jim

N2ADR

[Sid Boyce]

Hi Jim,
Thanks for the update which seem to be coming faster than I can complete
earlier boards.
1.41 looks like it would be a good place to put a stake in the ground
and complete a build.

For the 1.32 board, only today I received the SN10501D from Farnell, the
ULN2803 and NCP4623 from RS Components.
73 ... Sid.

On 30/03/16 17:14, James Ahlstrom wrote:
> Hello Group,
>
> I finally got my 1.40 boards back and I built the dual op amp buffer.

> Â Initial results look excellent. Â The output is 19.8 dBm at 7.2 MHz
> and 17.5 dBm at 29.7 MHz. Â This is with the 7-pole filter and the
> back termination. Â I tested the op amp with my tracking generator,

> and the -3 dB point is 64 MHz, so there is room for increasing the

> gain and output. Â I measured the output from the AD9866 alone, and it

> drops only 0.6 dB from 7.2 to 29.7 MHz, so the extra drop must be in
> the filter or the output transformer.
>

> Next I want to make new 1.41 boards for release. Â These will have a
> center pad for the OPA2677 op amp for greater power. Â And I am

> replacing the voltage regulator with a zener diode and emitter

> follower since the regulator is hard to source overseas. Â I will post

> more results once I order the new boards and make more measurements.
>
> Jim
> N2ADR

> --
> 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

[Alan Hopper]

Jim,

this sounds very promising.  Please excuse the following ramble just in case it is relevant to the front end design. I've been working on harmonic cancellation and have modified firmware that returns demanded dac values interleaved with the raw adc wideband data.  In trying to measure the delay between dac and adc, I made the mistake of testing at low frequency (<1MHz) so I didn't have to guess which multiple of wavelength the delay was in. This had me running around in circles because the current basic frontend appears to attenuate and phase shift frequencies below 3MHz ish so the apparent delay is not constant.  Is this just a fact of life or will your frontend work better at lower frequencies?

73 Alan M6NNB

[John Williams]

Jim,

This is indeed good to get back to 20dbm. I can then remove the 2n2222 preamp and give back some real estate. Is the output differential? I am toying with mounting a companion RD16HHF1 in the place of the 2n2222 to have a push pull amp. If it is not differential, I will plan on a transformer to make it so.

Great work!

As I mentioned before, if you want another set of eyes to take a look at the 1.40 board you have just tested, Joe WA9CGZ has some very nice test equipment and the skills to operate it very well.

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.

[James Ahlstrom]

Hello Allen,

I looked at the schematic, and I don't see a reason for phase shift until you get to the series blocking capacitors.  At 3 MHz, the reactance is 0.5 ohms, so I don't think that is the problem.  Maybe you could measure the phase shift with a scope.

The 1.4 board has a transformer, so at low frequencies that could be a problem.  But the 1.3 board lacks a transformer.

Jim

N2ADR

[James Ahlstrom]

Hello John,

The output of the 1.40 board is differential, and can be directly used by a subsequent push pull amp.  There is currently a transformer at the output to convert to single ended.  I have attached a schematic of the 1.41 board.  It is the same as the 1.40 board except for the voltage regulator.  If the following amp has a decent input impedance, the back termination resistors can be reduced, and you will get an additional 5 dB.  And I plan to increase the gain.  Eventually, the power capability of the op amp will be limiting.

Jim

N2ADR

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

Hello Jim,
The new front end board looks very good, glad to see the Hermes opamp in there - its a good choice!
What output transformer did you use in the end? With my Hermes I changed it from a 1.5:1 to a 1:1, reduced the supply to 9V and got ~+24dBm with good linearity.
One point to note that people did blow the output stage occasionally due to a poor antenna VSWR I assume. With mine wound back it should be more robust.
Would you have room to squeeze in an Rx anti-alias filter using smd L's and C's, say a 5 or 7 pole elliptic, it would make it more complete?
Could come up with values if required.
Thanks,
Andrew
G4XZL

[Steve Haynal]

Hi Jim,

This looks great! Nice power output, proven op amp, cleaner output than 1-2 transistor-only design, and can support differential output.

73,

Steve

KF7O

[Steve Haynal]

Hi John,

I hope that in the V2 layout, a builder will have the option of leaving ogf the transformer on V2 and feeding a differential signal directly to the companion card. This might only work well in the sandwich configuration where the distance is very small between the Hermes-Lite and the companion card.

73,

Steve

KF7O

[John Williams]

Yes, I was wondering the best way to feed the differential signal between boards...

[James Ahlstrom]

Hello Andrew,

On Wednesday, March 30, 2016 at 4:26:50 PM UTC-4, irbsu...@yahoo.co.uk wrote:

What output transformer did you use in the end?

I am using a 1:1 homebrew transformer on a BN-43-2402 core, six turns each winding.

 

With my Hermes I changed it from a 1.5:1 to a 1:1, reduced the supply to 9V and got ~+24dBm with good linearity.

That is good to know.  Are there any resistors at the output of the op amps?  Or do you connect directly to the transformer?

 

One point to note that people did blow the output stage occasionally due to a poor antenna VSWR I assume.

Yes, I was worried about that.

 

Would you have room to squeeze in an Rx anti-alias filter using smd L's and C's, say a 5 or 7 pole elliptic, it would make it more complete?

There is no room.  But I could possibly fit it if I used a small SMD transformer at the Rx input, and eliminated the Tx/Rx switch.  Is anyone using the Tx/Rx switch?

Jim

N2ADR 

[James Ahlstrom]

Hi Steve,

The transformer is homebrew on a BN-43-2402 core.  The hole spacing is on 0.1 inch centers.  So you can mount a pin header in place of the transformer.

While we are on the subject, the Minicircuits X65 and the MaCom MABAES0060 both have a max power of 250 mW, or 24 dBm.  So I don't think they are suitable unless we design for lower power.  No manufacturer will wind the transformer by hand, so the transformer is a problem for completed boards.

Jim

N2ADR

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

Hi Jim,
Yes, I retained the 4.7R resistors as per the original Hermes. If its possible to avoid Minicircuits parts that would be a help as they are difficult and expensive to buy (in UK) in small volumes.
Hand wound with a dual footprint for a pre-wound part is a good compromise. CX2074NL or CX2147 or MACOM MABA series all from Digikey might be worth looking at, although noted about the power handling -  a common problem unfortunately.
Okay about the receive filter, it could be quite small.
I would be interested to look at the pcb design files, would that be possible?
Thanks,
Andrew
G4XZL

[James Ahlstrom]

Hello Andrew,

Yes, you are welcome to the PCB design files.  Do you want the whole Kicad project, or the Gerbers?  I no longer have files for the 1.40 board as I am already working on 1.41, the version without the voltage regulator.  But I can send you 1.41.  If you just want to look at the receive circuit, there is no change from the 1.3 board, and the 1.3 files are all on github under softerhardware.

Jim

N2ADR

[John Williams]

You know that boards are dirt cheap on Elecrow, Jim. You could expand the board size to accommodate the RX filter and not add any cost. The increments are 5cmx5cm then move to 10cmx10cm. The boards on Oshpark are another pricing schema at $5 per square inch. So I would not let Oshpark pricing drive your decision here on space.

If this ends up being the final 1.X frontend it would be really good to have the RX anti-aliasing filter. Andrew and I are contemplating removing the RX filter from the PA board.

John

[James Ahlstrom]

Hello John,

Not increasing the board size was not based on cost.  I was afraid the bigger boards would not fit in the space people had available.  For example, if I increase my board opposite the fingers by 10 mm, it hits some screws.

Let me work on the Rx filter.  I can cut back on the size of the SMA connector footprint and the T1 footprint, and make something fit.  I guess we want a 5-pole 1dB Chebyshev at 32 MHz, right?

Jim

N2ADR

[John Williams]

I will let Andrew suggest a filter. I was thinking of suggesting making it wider, actually.

[Heikki Ahola]

Hello group!

I was just wondering if any decision or proposal has ever been made regarding the cabinet size for V1 or V2. Personally I have available all the space that is needed plus some extra. In the past I have started too many projects with a cabinet far too small for the purpose and I have learned my lesson ( see my contribution in the Gallery). I would imagine that as long as we are dealing with an experimental project/device, the physical dimensions should not be an issue. It is not feasible to try to squeeze the components in a pre-defined  space only to find the assembly much harder and the performance to suffer. 

73 de Heikki (OH2LZI)

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

Hi Jim,
Regarding the filter, basically it needs to supplement the internal lowpass.
So with a 73.728MHz clock we need to achieve a good stopband by ~36.8MHz, so yes a cut-off around 32MHz sounds about right, best if its an elliptic.
Go for a 5 pole elliptic using 0805 SMD inductors and caps. I'll give you some values early next week, may struggle with single values though, normally needs paralleled caps to get the right value.
Regards,
Andrew
G4XZL

[James Ahlstrom]

Hello Andrew,

I usually use Elliptic filters for Tx lowpass, and Chebyshev for reconstruction and anti-allias filters.  The Chebyshev has much better stop band at higher frequencies.  Could a Chebyshev be used?  It has fewer parts, and I am struggling for space.

Jim

N2ADR

[James Ahlstrom]

Hello Andrew,

Now I see what you mean.  The elliptic produces greater attenuation at 36.8 MHz, and the internal AD9866 lowpass handles higher frequencies.  So an elliptic is best.

I will find room for a 5-pole elliptic filter.  Parallel capacitors can be accommodated by soldering them on top of each other.  Let me know the values when you get a chance.

Jim

N2ADR

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

Hi Jim,
If its possible to send me the Kicad files for 1.41 that would be preferable, easier to view than the gerbers.
Thanks,
Andrew

[James Ahlstrom]

Hello Group,

Here are some design notes and measurements of the 1.40 frontend board.  I increased the voltage gain from 9.05 to 11.65.  The amplitude at each op amp output is now 7.40 Vpp.  With two 22 ohm back termination resistors, the output is 22.20 dBm at 7.2 MHz, and 19.70 dBm at 29.7 MHz.  If I change the back termination from 22 to 4.7 ohms, the output is 26.0 dBm at 7.2 MHz and 22.8 dBm at 29.7 MHz.  I think this is close to our final result.

I plan to use an 11 volt zener followed by an emitter follower to power the op amp.  This is a well known circuit, and it has the property that if the supply voltage goes below 11 volts, it still powers the op amp with a 1 volt drop across the transistor.  That could be useful if the Hermes-Lite is powered from a battery.  With an input supply voltage from 11 to 16 volts, the op amp supply is 10 to 10.3 volts.  This protects the op amp from over voltage, and although the regulation is poor compared to a voltage regulator, the op amp does not need regulation due to its power supply rejection.

With 10 volts Vss available, and 2.5 volts head room, we get 7.5 volts peak to peak at the output of each op amp as the design goal.  This equates to 27.5 dBm into 50 ohms push pull.  The back termination reduces this by up to 6 dB.  And there will be losses in the transformer.  The power dissipation in the op amp is up to 528 mW.  But if a low impedance load is attached, the current will increase and the maximum power dissipation of the op amp will be exceeded.  I plan to test an opa2677 which has a power pad package.

I am finishing up the 1.41 boards, and adding an elliptic receive filter.  If they test good, maybe we are done with front end boards.

Jim

N2ADR

[ZL2APV]

Hello Jim,

This has really got somewhere now. I have lost track of what spurs are a problem with the new frontend and would be grateful if you could publish the frequencies that you would like a filter to attenuate. The other thing to bear in mind is that with a 7 pole elliptic there two notches which can be manipulated to an extent although other characteristics will be compromised. It may be best to use a 7 pole if useful out of band attenuation can be obtained. Of course that may be too big a problem in terms of board real estate. The other issue is that at 30 MHz regardless of 5 or 7 pole there is some advantage in using T30 cores for improved Q meaning less in band loss plus better corner frequency response and with the low Rx power T30-2 will do the job at less than 8mm diameter. Typically less than 10 turns will be needed so winding will not be a big issue leaving only pruning the inductance as a hurdle and again that can be overcome using the Hermes-Lite plus 4 or 5 components as the only instrument needed. Even without pruning the filter will be not too bad although the notches may not come out exactly where desired but will still be helpful.

73, Graeme ZL2APV

[ZL2APV]

Whoops! that is what comes of listening to a contest and typing at the same time. My references were to T30-10 cores not T30-2 cores - sorry.

Graeme

[James Ahlstrom]

Hello Andrew,

Please send your email address to jahlstr at gmail dot com.

Jim

N2ADR

[James Ahlstrom]

Hello Group,

Here is the latest schematic for 1.41.  The elliptic Rx filter values are only place holders until Andrew can provide real filter values.

Jim

N2ADR

[Steve Dick]

Hi Jim and group. You might want to consider a “capacitor multiplier” circuit which is a simple mod to the zener based voltage regulator.  For example, see: http://www.edn.com/electronics-blogs/the-practicing-instrumentation-engineer/4422750/Simple-circuits-reduce-regulator-noise-floor

 

Regards. “Digital Steve”, K1RF

--
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.

Virus-free. www.avast.com

[in3otd]

Hello Jim,
I finally found some time to build and test the TX part of your frontend boards. I have first tested the v1.32 - no surprises here, it works exactly as you described, hi. As usual, all the details are on my website, see http://www.qsl.net/in3otd/ham_radio/Hermes-Lite/frontend_v1.3.html .

I have also build the v1.4 dual opamp circuit using the v1.32 PCB; it took some time (and patience) to fit the new amplifier on that PCB but it seems to work quite well. Schematic should be as 1.41, except R7 and R8 which are 22 ohm instead of 24, but should not make a lot of difference. Partial results are at http://www.qsl.net/in3otd/ham_radio/Hermes-Lite/frontend_v1.4.html - will add more data in the next days.

As you already saw, the frontend v1.4 performances are quite good and distortion is also very low, especially at low power levels. To be really nit-picking I could say that the frequency response drops a bit too much at the end of the HF (similarly to v1.3) but I do not have a good understanding of the reasons for that. My simulations show that the filter may drop 1 dB overall, not sure if the other missing dB is due to the opamp, the TxDAC or other things.

73 de Claudio, IN3OTD / DK1CG

[Steve Haynal]

Hi Heikki,

There are two target enclosures listed on v2 wiki. Both target 100mm by 80mm boards. The taller enclosure should support the "sandwich" configuration of the Hermes-Lite V2 board and the companion filter PA board. John has managed to keep his filter/PA board within the 100mm by 80mm bounds. That is what I am shooting for with v2 also. I think it is good to have a target, but if it turns out to be impossible or more trouble than it is worth, we can increase the size. For price breaks on PCB fabrication, it would be good to then not go over 100mm by 100mm.

73,

Steve

KF7O

[Steve Haynal]

Hi Claudio,

Thanks again for your well-documented analysis. I see two broken links to graphs under the 

Output vs. TX drive level

section of the v1.4 page.

73,

Steve

KF7O

[in3otd]

Hello Steve,
yes, that is part of the "will add more data in the next days"; those measurements were ongoing overnight, will postprocess them and update the site "soon". I should have removed the whole section to avoid confusion, after copy/pasting from the v1.3 page :) .

73 de Claudio, IN3OTD / DK1CG

[James Ahlstrom]

Hello Claudio,

Thanks for taking a look at this and providing more measurements.  I also think the output drops too much at 30 MHz and I also do not have a good understanding.  I did measure the output of the AD9866 at the 33 ohm resistors without a filter in place.  The output is -16.6 dB at 7.2 MHz and -17.2 at 29.7 MHz, so 0.6 dB is in the AD9866.  The measurement was made with an active probe and my spectrum analyzer.  A scope is not adequate for this measurement.

The filter does drop at 30 MHz.  I wonder if there is significant added capacitance from the traces between the AD9866 and the filter.  When V2 is available, we need to check the filter response and adjust the component values for the final board parasitic capacitance.

The OPA2674 output drops as power is increased.  As R6 changes from 82 to 47 ohms, the difference between 7.2 and 29.7 MHz power changes from -2.31 dB to -2.71 dB.

There could also be a problem with the transformer response.  The drop across HF is probably the result of multiple small losses.

Jim

N2ADR

[James Ahlstrom]

Hello Group,

I would like to order 1.41 boards within the next few days.  It takes some time for them to come.  We can still change component values.  Does anyone have changes other than component values?

Jim

N2ADR

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

Hi Jim, Graeme,
Currently I am designing the filter to use 0603 or 0805 SMD inductors, I really don't need toroids in the design but if people really want them can you put in a dual footprint please?
If anybody wants to use hand wound toroids then they will definitely need to be able to measure the inductance or be able to measure the response of the filter otherwise there is no chance of getting the notches in the correct place in my experience, they are just too variable.
Regards,
Andrew
G4XZL

[in3otd]

Hello Jim,
which inductors did you use for the LPF before the OPA2674? I used the Murata LQW2BASR24J00 and LQW2BASR27J00 and I have just re-run a simulation using the spice model available on the Murata website: it turns out that the inductors Q is worse than I remembered, around 30 or so at 30 MHz, so the actual drop due to the filter alone is around 1.5 dB at 30 Mhz.  This plus the 0.6 dB drop from the TxDAC you measured is not far from the 2.5 dB drop I measured on the frontends v1.3/v1.4.
Slightly bigger inductors, like the LQW2UASR27J00L or CW252016-R27J have a slightly higher Q. but not all values are available. There are even bigger ones from other manufacturers, still wire-wound on ceramic core, which have a lower DC resistance but no models are provided for simulation.

Is there a particular reason to have the 390 ohm termination resistor (R1) on v1.4? By removing it we can recover about 1 dB and by adding some resistor on the filter internal nodes the filter gain could be equalized. A simulation removing R1, adding 1 kohm in parallel with C2 and C3 and changing C4 to 100 pF gives a filter with a slightly upward slope, may be helpful to compensate for the TxDAC drop.

73 de Claudio, IN3OTD / DK1CG

[ZL2APV]

Hi Group,

Yes Claudio the loss conclusions you reached were the same as mine which is why I raised the possibility of using toroids however the point that Andrew made is right on the mark as toroids can come out at up to 10% away from calculated value. Another issue to consider is the capacitors in the filter as non RF types can have Q in the low hundreds instead of around 1 to 2000. The filter error is particularly noticeable in the corner area which of course shows up more at 30 MHz. To get a reasonable filter you need inductor Q's of at least 80.

73, Graeme ZL2APV

[rodwall1]

Hi Graeme and all,

Graeme, you bring up a important point with regard to component Q.

How do we control the quality of the components. When builders start building using parts from their junk box.

One way is to specify a component make and part number on the BOM. And add a note indicating that any substitution must have specifications as good as the substituted part.

We don't want questions as to builders asking, why their filters are not working. This may reduce these type of questions.

Roderick Wall, vk3yc.

--

[Steve Haynal]

Hi List,

One interesting way to see the effect of reduced Q on filters is to try out Coilcraft's LC Filter Designer. This is quick and easy to install, and it uses models of real components. As you reduce the physical size of the inductors and Q deteriorates, you can see the effects on the filter.

Perhaps there is a way to use the Hermes-Lite before installation of final inductors to measure individual inductors. This way, builders can roll their own but have a way to adjust the inductors properly.

I'm actually a bit hesitant to sell complete transmitters without proper FCC approval as I think that might violate some rules. It may be safer to leave at least the inductors off and require a builder to finish those so that it is still a kit.

73,

Steve

KF7O

 

On Monday, April 4, 2016 at 2:40:37 PM UTC-7, Rod Wall wrote:

Hi Graeme and all,

Graeme, you bring up a important point with regard to component Q.

How do we control the quality of the components. When builders start building using parts from their junk box.

One way is to specify a component make and part number on the BOM. And add a note indicating that any substitution must have specifications as good as the substituted part.

We don't want questions as to builders asking, why their filters are not working. This may reduce these type of questions.

Roderick Wall, vk3yc.

[James Ahlstrom]

Hello Claudio,

Yes, I used the same Murata LQW2BASR24J00 and LQW2BASR27J00 inductors.  The Q of 30  at 30 MHz compares to the advertised Q of 44 at 250 MHz.

The 390 ohm resistor is there to provide a known output impedance for the filter.  The filter was originally designed for v1.3 which had an input impedance of 400 ohms.  So the filter is designed for 66 ohms in, 400 ohms out.  The 390 ohms swamps the input impedance of the op amp at the output of the filter.

Tweaking values to achieve an upward slope sounds like a good idea to me.   But I don't understand the logic of the 1K resistors.  I would like to keep the filter as close to a Chebyshev as possible.  Remember that the real point of the filter is to suppress output at VHF and UHF, and suppress that darned spur at 45 MHz on 10 meters.

I am not too concerned about a 3 dB drop at 30 MHz.  Every amp I ever built has that, and so will the Tx amps that follow the Hermes-Lite.  I am happy flattening the gain in software, where we have 7.5 dB available.

Jim

N2ADR

[in3otd]

Hello Jim,
I meant that in theory the filter could be designed to work unterminated (loaded only by the 4400 ohm of the bias network, which is practically an open circuit compared to the 66 ohm of the source) so that there will be no losses in the termination and a little gain could be recovered.
Then, since we now have a little more gain, we can spend it to "predistort" the filter response, i.e. introducing more losses at low frequency (via the 1 kohm resistors), so that the overall response looks flatter. The response in the stopband is practically unchanged.

I have done a quick measurement on the frontend v1.4 with the LPF modified as described previously, results are shown in the graph below:

the response is a bit better, but the simulation was showing even better results (as usual...). Probably also the TxDAC and opamp loading should also be taken into account to compute the overall response.

73 de Claudio, IN3OTD / DK1CG

[James Ahlstrom]

Hello Claudio,

Here are some more measurements of the IMD for the 1.4 board.  The gain resistor R6  is changed from 82 to 62 ohms for a gain of 11.65, and the back termination is 4.7 ohms.  I am using Quisk with the IMD mode to generate the test signal.  So the figures include everything from the AD9866 to the antenna.  At 7.2 MHz, the Spot feature gives an output of 25.84 dBm, while each tone of the two-tone test is 19.79 dBm.  This is -6 dB as expected, since the Spot level is the PEP.

F MHz      One Tone dBm      IMD3 dBm        IMD5 dBm       IMD7 dBm

  7.2               19.8                      -44

 21.3              18.6                      -29                  -42

 29.7              16.7                      -26                  -38                   -45

You stated that for an IMD3 of -30 dBc the output power for your configuration is 60 mW per tone.  That would correspond to a Spot level (and PEP level) of 240 mW.  This is close to your 1 dB compression level of 220 mW.  So it looks like the op amp provides low IMD until it clips.

My gain produces an op amp output voltage of 7.4 Vpp.  This is near the clip level when powered by 10 volts.  So I also conclude that the op amp provides low IMD until it clips.  Then distortion rises rapidly.  So we should design for a maximum output slightly below clip.  Since the AD9866 produces a precise output voltage, this fixes the gain given the supply voltage.  Then we just have to decide the back termination resistors.  The 24 ohms was chosen to produce a 50 ohm output impedance.  Lower values produce more output, but don't provide protection for the op amp in case of a low impedance load.

Jim

N2ADR

[James Ahlstrom]

Hello Claudio,

On Tuesday, April 5, 2016 at 2:55:52 PM UTC-4, in3otd wrote:

I meant that in theory the filter could be designed to work unterminated (loaded only by the 4400 ohm of the bias network, which is practically an open circuit compared to the 66 ohm of the source) so that there will be no losses in the termination and a little gain could be recovered.

That is fine with me, but I am not sure it is necessary.  I only added the 390 ohm resistor because I did not know what the input of the op amp looked like, and the 390 was surely less.   Recovering gain is not important as R6 adjusts the gain.  Your gain (at 82 ohms) is 19.13 dB, and mine (at 62 ohms) is 21.33 dB.

Then, since we now have a little more gain, we can spend it to "predistort" the filter response, i.e. introducing more losses at low frequency (via the 1 kohm resistors), so that the overall response looks flatter. The response in the stopband is practically unchanged.

I will have to defer to you on this.  It sounds like a good idea though.  To provide for this on the 1.41 board we will have to solder the resistors on top of the capacitors; not a problem.

Jim

N2ADR

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

Hi Jim,
Unfortunately having done some simulations a 5 pole elliptic isn't going to give enough rejection to be that useful even when cascaded with the internal LPF. It really needs to be a 7 pole elliptic. I can see now that you really are struggling for space on the pcb and that has meant that the tracking of the LPF is rather awkward and the groundplane is rather fragmented. On reflection, it might be best to leave it off this one and look at adding it to the next? This would allow you to route the other tracks more easily and recover the groundplane?
If you do keep the filter it would be worth adding some more ground vias especially around the filter.
How do people feel about 0603 components, I would say they are no more difficult to solder than some of the other parts e.g. RF switch?

Andrew
G4XZL

[Steve Haynal]

Hi Group,

I wonder if it would be better to only use a 3-pole filter on the balanced input of the op-amps and then a 7-pole unbalanced after the transformer. The advantages I see to this are:

• The unbalanced filter requires fewer inductors and hence less board space. This implies fewer inductors to wind for those going the wind your own route. Also, fewer inductors implies it is easier to use higher Q inductors for a better filter.
• The unbalanced filter can be shared with RX when using the TR switch.
• Easier to achieve less imbalance in a smaller balanced filter.

Regarding 0603, I prefer to not to use them in V2 unless forced to do so for area reasons. I think capacitor ESR is lower and Q higher for 0805, more pad space to touch test probes to, easier to hand assemble and repair. We will have to see exactly what the area constraints look like.

73,

Steve

KF7O

[Alan Hopper]

Jim,

thanks for looking at this, sorry for my slow reply. I had scope trouble, first my aged hp1703a analog storage scope died, then I discovered the bargan Rigols are out of stock everywhere, in desperation I tried the bescope I got free with my bemicro cv and learnt not to trust it, finally I got the hp going again.  On my basic front end with a 16db  pad between tx and rx I see the phase shift and attenuation on the scope, it is 3db down at about 800khz.  With the pad not connected to the rx input and terminated I don't see the effect so guess it is the rx input circuit.  I'm not suggesting this is a real problem but it caught me out so I thought I'd mention it. 

Thanks again

73 Alan M6NNB

On Wednesday, March 30, 2016 at 8:22:24 PM UTC+1, James Ahlstrom wrote:

Hello Allen,

I looked at the schematic, and I don't see a reason for phase shift until you get to the series blocking capacitors.  At 3 MHz, the reactance is 0.5 ohms, so I don't think that is the problem.  Maybe you could measure the phase shift with a scope.

The 1.4 board has a transformer, so at low frequencies that could be a problem.  But the 1.3 board lacks a transformer.

Jim

N2ADR

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

Hi Jim,
I've had another look at the alias filter and actually the worst signals to reject (in the UK at least) are Band 2 FM which is 88-108MHz, very large here. So a 5 pole Cheby (not elliptic) would be useful.
Here are some values if you decide to keep the filter (C18 & C19 can definitely go if it helps the layout), up to you:
C11=100pF C0G
C16=180pF C0G
C17=100pF C0G
L4=330nH Coilcraft 0805CS or similar (may need a slightly larger footprint than a standard 0805)
L5=330nH Coilcraft 0805CS or similar (may need a slightly larger footprint than a standard 0805)
This gives about 50dB rejection at 90MHz plus about 38dB from the internal filter.
Won't do much for the closer alias responses but actually there aren't many signals in the second nyquist band these days so might be able to get away with it?
Hope that helps.

Also having a look at the TXDAC balanced output filter, Claudio's idea of a singly terminated filter is a good one, can give you some values to try.
Looking at Claudio's results the current one seems to have quite a bit of ripple and rolls off quicker than one would expect.
Andrew
G4XZL

[James Ahlstrom]

Hello Andrew,

Thanks for the new filter.  I will keep all the pads for the elliptic in place.  So we can build your filter, and also allow the construction of the elliptic if someone want to try it, or if they discover troublesome signals in a different frequency range.  I think the ground plane is quite good, and eliminating pads for C18 and C19 is not much of an improvement.

While we are on the subject, I always leave my transmit low pass filters in line on both transmit and receive.  Since they must handle power, they must be low loss, and so they don't add much to the receive noise figure.  So the role of the Rx filter we are discussing is to remove troublesome images that are missed by the Tx filters plus the RC filter internal to the AD9866.

By the way, your UK Band 2 FM is the same in the USA, and just as troublesome.

Jim

N2ADR

[James Ahlstrom]

Hi Steve,

I don't like the 3-pole before the op amp.  The op amp has gain to several hundred MHz, and we can't be sure that the strong 2Fs +/- Ftx images (and others) are not going to produce spurs.  I also don't think that winding your own toroids for the filter is an option.  It is just too hard to measure these small inductances.  The total cost of a 7-pole balanced Chebyshev filter is $1.90 quantity one from Digikey, and I don't think it is worth trying to share a filter.

Jim

N2ADR

[Steve Haynal]

Hi Jim,

I must disagree just a bit. Figure 22 in the ADA4311 datasheet recommends a 3-pole balanced before the amp. Filters later in the gain stage work pretty well as evidenced by clean output from my early version of John's PA and QRPlabs filters. The SoftRock 6m/4m/2m receiver uses wind your own 350 nH inductors. You can still have a good number of turns to play with if the toroid diameter is small and permeability low (material 10 or 17). Plus, we should be able to use the Hermes-Lite to tune inductors as part of the build process. Finally, you have the higher Q and lower loss, and now questions about selling a TX that is a kit or not.

But what you are pursuing is also very reasonable and has different advantages, and since you are doing all the work, I will gladly borrow your work. I am considering the following changes/additions though:

• An option for a small daughter board for a completely different frontend. This would fit as a sandwich on the opposite side of the PA/filter companion card, connect to the analog outputs of the AD9866 via standard arduino-like headers, and then feed the balanced and/or unbalanced output of the onboard frontend. The onboard frontend would have to be removed if this option is in place - both could not coexist. I still want to experiment with the TH6226! The connection holes can be thought of and used as test points when no alternate frontend is in place.
• Option for multiple receive antennas. One without filter for 6M use, one without filter for resistor tap measurements as discussed in the past. I would like to use the Peregrine device, or something similar, provided it doesn't introduce noise. Depending on space, these options may only appear on an optional frontend. 

73,

Steve

KF7O

[James Ahlstrom]

Hello Steve,

I don't think I understand what you are after here.  I need to see a schematic when one becomes available.  But right now my time is best spent in making the 1.41 boards, testing them, and publishing them.

The 1.41 boards differ from the 1.40 boards as follows:

1.  I replaced the voltage regulator with a zener and emitter follower because it is hard to get the regulator overseas.

2.  Because we are designing for higher power, I changed the footprint of the op amp to accommodate an OPA2677 with a center pad for a heat sink.

3.  I added the 5-pole Rx filter that Andrew requested.  This can be a Chebyshev or ellipical.

I intend this board to be available as an alternative to the 1.32 board for those who want 20 dBm instead of 10 dBm, or want the other features.  I ordered the boards.  If they test good I will send you the project frontend/v1.4.

Jim

N2ADR

[Steve Haynal]

Hi Jim,

Agreed that your work on 1.41 is time well spent. The ideas I have are from the v1.25 frontend card and deal with stuff outside of the core amp and filters you are working on. Hopefully I can get them down on a schematic this week or next.

73,

Steve

KF7O

[James Ahlstrom]

Hello Group,

Winding transformers sometimes seems to be a black art, and I often just copy transformers that others have published.  But I want to share some things I learned when I redesigned the output transformer for the 1.41 front end board.  I measured a decreased transformer response of -0.93 dB at 29.7 MHz relative to 7.2 MHz, and I wanted to flatten the response.  The total change in output when changing from 7.2 to 29.7 MHz is -3.16 dB.  I am using 4.7 ohm back termination resistors.  Output at 7.2 MHz is 26 dBm, or 400 mW.

The original transformer was 6 turns for each winding on a BN-43-2402 core.  It showed a -3 dB output at 50 kHz, so it seemed I could favor the high frequency side by using a lower inductance.  So I tried a second transformer that was four turns bifilar on a BN-61-2402 core.  This only has 2.4 uH inductance, or 28.7 ohms at 1.9 MHz.  But the response at 29.7 MHz relative to 7.2 MHz changed from -3.16 to -2.37 dB, a good improvement.  The performance at low frequencies seemed too good to be true, and it was.  Although the output at 1.0 MHz was good, the supply current increased from 103 ma to 168 ma and the termination resistors were overheating.  There wasn't enough impedance and thus the excessive current.

I needed more impedance.  The third transformer has 4 turns bifilar on a BN-43-2402 core.  This has 20.4 uH, or 231 ohms at 1.8 MHz.  This showed no excessive supply current down to 1.9 MHz, and response at 29.7 MHz relative to 7.2 MHz changed from -3.16 to -2.36 dB, the same as the second transformer.  The first transformer used the same core, had six turns instead of four, and was wound 6 turns and 6 turns instead of bifilar.  The third was much better even though it used the same type 43 ferrite.  So transformer three is the final design.  But it will not work at the new 630 meter allocation; you will need a different transformer.  Maybe we could increase the turns, but I didn't try that.

So the lesson I learned is don't be afraid to use higher permeability materials for transformers; think of using bifilar windings when possible; and check the supply current, not just the output when testing.

If you don't have any type 43 cores, you can use a type 61 core and increase the number of windings to 11 or 12.  If you can't fit that many windings, use fewer and check the performance at low frequencies (if you want to work 160 meters).

Jim

N2ADR

[John Williams]

Jim,

I can send you a BN 73-2402 core if you feel like continuing the experiments.

John W9JSW

[James Ahlstrom]

Hello John,

I guess the 73 core would be for bands below 500 kHz.  I don't want to work on this now because I am still studying power dissipation issues for 1.41.

Jim

N2ADR

[John Williams]

No problem, Jim.

Just wanted to offer one to you if you were so inclined. Yes, it is for the lower bands. I can have Joe WA9CGZ test with it when we release the 1.41 boards. He is doing work on those bands.

John

[James Ahlstrom]

Hello Group,

The 1.41 board runs at a voltage gain of 11.65.  The optimal feedback resistor for the OPA2674 is 240 ohms, but we are using 330.  That will reduce the bandwidth.  To see if that matters, I changed the feedback resistors to 270 and the gain resistor to 51 ohms for a gain of 11.59.  I tested at high power, 26 dBm.  It didn't make any difference.  So the 330 ohm feedback resistors are fine for a design up to 30 MHz.

Using higher feedback resistors decreases bandwidth and increases stability.  We don't want bandwidth beyond 30 MHz.

Jim

N2ADR

[James Ahlstrom]

Hello Claudio,

I finished the power design.  The only thing interesting is that we need 1/4 or 1/2 Watt resistors for the back termination.  Panasonic makes a 1/2 Watt in 0806.

Then I had time to test your filter, and it works great!  After reducing the gain from 11.65 to 9.80 and using 24 ohm back termination resistors, here is the output:

F MHz        1.9       3.8      7.2      14.2       21.3      24.9       29.7

Out dBm    21.25   21.03   20.77   21.51    20.57    21.13      19.80

I will put your values in the BoM and change the schematic.  But there are no pads on the board for the 1k resistors, so everyone will have to solder them on top of the capacitors.  We may get back some of the output at 29.7 MHz from the higher clock in V2 if it comes from the AD9866.

Jim

N2ADR

[in3otd]

Hello Jim,
to output power flatness looks quite good, quite better than what I measured, hi.
I think the filter bandwidth may be a little too close to 30 MHz, I wanted to optimize the filter further and seen if the drop at 30 MHz could be reduced without compromising the rejection at the images but did not have the time yet.
Which transformer did you use for the results below, 4 turns on a BN-43-2402 ?

73 de Claudio, IN3OTD / DK1CG

[James Ahlstrom]

Hello Claudio,

Yes, the transformer is four turns bifilar on a BN-43-2402.  I think using bifilar twisted pair enameled wire is important.

Jim

N2ADR

[James Ahlstrom]

Hello Group,

I got my 1.41 boards back and have started testing them.  I plan to add Claudio's 1000 ohm resistor pads to the board design, and release it as version 1.42 as soon as 1.41 tests good.  The first users will be the testers.  The new filter by Claudio really woks well, and I want to publish a proper board for it.

Jim

N2ADR

[James Ahlstrom]

Hello Group,

I tested the 1.41 boards with an OPA2677 and 24 ohm back termination resistors with Quisk "Spot" and got good results.

F  MHz              1.9         3.8       7.2       14.2      21.3     24.94      29.7

Output dBm    21.15     20.93    20.68    21.62    20.70   21.25     20.52

Harmonics at 7.2 MHz in dBm are:   20.2    -29.5    -41   -37

IMD at 14.2 MHz is less that 60 dB below one tone.

There is no visible (-60dBc) spur at 44 MHz when transmitting on 10 meters.

The only surprise is "high" harmonic output on 10 meters of -40 dBc which should be suppressed by the filter.

The gain flatness is quite surprising to me.  Claudio's filter is doing a great job.  I am therefore adding pads for the 1k resistors, and publishing the changed board as version 1.42.  It is shared on OSHPark, and will be available in github.  Please refer to the frontend.html documentation in the 1.4 directory.

The OPA2677 is the version with the pad on the bottom that must be soldered to the board.  There is a hole for this purpose.  The part number is OPA2677IDDA.  I don't see a reason to prefer the OPA2674 over this part.

The maximum output voltage available is 5.3 volts RMS.  In 50 ohms, the maximum power is 27 dBm, 6 dB below our power, as expected.  To get more power, reduce the 24 ohm back termination resistors.  I have tested at 4.7 ohms each.  See frontend.html for cautionary statements.

Jim

N2ADR

[Steve Haynal]

Hi Group,

Jim's v1.42 is now on github master and v1_22. 

Thanks and 73,

Steve

KF7O

[John Williams]

Like I did for V1.32, I will order a set of these from Elecrow today and make them available to other builders at cost.

John - W9JSW

--

[John Williams]

20 boards are ordered. 4 are spoken for so far.

[Alan Hopper]

John

1 please
Alan 2E0NNB

[Glenn P]

Include me also please.

I have been absent from doing any smd work here for a mth or so and for a bit longer due to cataract removal on both eyes. But by the time the boards are ready and here, i should be ok to go.

The Op-amp mentioned though is not available here.  Is it possible to get one also?

glenn

[pascal.v...@gmail.com]

Hi John,

I am interested by one board also

Thanks

[Wolfgang DL9UFB]

Hello John,

I would also like a board.

Thanks Wolfgang

[John Williams]

Jim,

Attached is a new 1.42 BOM with Digi-key part numbers added...

John

On 4/22/2016 6:05 AM, John Williams wrote:

[Ted Pater]

Hi John,

One board for me please.

Ted  VE7VIB

[David Rutledge]

Hi John,

Please add me to the list for a 1.42 board. I'm new to the list and I thought I posted a reply earlier but I didn't see it post.  Let me know what I owe you for the board and if Paypal is OK.

73,

David A. Rutledge

AL5M

[John Williams]

ok, David. Boards are on the way. Should have them this week.

Will respond with paypal info when ready to ship.

John

[John Williams]

Elecrow sent me 28 boards for an order of 20. 14 boards are spoken for so we have ample quantity for anyone else. Will be sending them out on Monday and Tuesday. Board cost will be $3.00 USD.

I have a BOM loaded on Digikey but do not know how to share it. The BOM id is 3478225. Updated BOM csv file attached.

Will notify shipping costs on an individual basis after they are on the way, then you can paypal me.

John - W9JSW

On 4/22/2016 6:05 AM, John Williams wrote:

[Steve Haynal]

Hi Group,

Jim kindly sent me a built-up instance of his v1.41 board. I have been very impressed and happy so far. Great job Jim! I haven't done extensive testing, but here are a few observations.

Receive is working well. I have been running WSPR with it this past week, and I consistently have more spots per band than others, unless the other person lives out in the country and has a huge antenna farm. Although my attic dipole is only 20/15/10M, I run WSPR RX also on 40/30/17M and do well. I run at LNA +22 dB gain with the RFAGC on. With the RX filter on this board, I do not see the strong images of FM stations in the 22-28 MHz range. For a stong local AM station with a steady carrier, I see a slight signal strength improvement (~1dB) for the new board when comparing head-to-head with an old frontend card.

I ran one spectrum analysis plot of TX in the 20M WSPR band shown below. I think it looks pretty good. You can see the harmonics. This is without any additional filtering, only the reconstruction filter on the new board. Also remember that my scope/spectrum analyzer setup generously gives me "bonus" spurs, such as the ones near the carrier and at 125MHz. 

I did transmit on 20M and it was nice to be spotted in New Zealand with only 100 mW of power. I was hoping that the change in power profile due to using the TxDAC might improve the drift with the Si510, but you can see that I still have drift on TX. Remember that V2 uses a completely different clocking structure that should have much less drift.

2016-04-30 15:30  KF7O  14.097061  -23  -2  CN85oh  0.1  ZL1RS  RF64vs  10986  228  

2016-04-30 15:30  KF7O  14.097114  -23  -2  CN85oh  0.1  KI7MT  DN46bs  857  75  

2016-04-30 15:30  KF7O  14.097047  -26  -2  CN85oh  0.1  K7GXB  DM34sn  1482  140  

2016-04-30 15:24  KF7O  14.097061  -20  -2  CN85oh  0.1  ZL1RS  RF64vs  10986  228  

2016-04-30 15:24  KF7O  14.097133  -13  -4  CN85oh  0.1  KK6DJC  DM12kw  1459  159  

2016-04-30 15:18  KF7O  14.097074  -23  -3  CN85oh  0.1  W0ANM  EN35ee  2271  80  

2016-04-30 15:18  KF7O  14.097069  -26  -4  CN85oh  0.1  KC9YGI  EN60gb  2922  89  

2016-04-30 15:18  KF7O  14.097025  -28  -3  CN85oh  0.1  N5CEY  EL16gq  3055  124  

2016-04-30 15:18  KF7O  14.097064  -18  -3  CN85oh  0.1  ZL1RS  RF64vs  10986  228  

2016-04-30 15:18  KF7O  14.097135  -14  -4  CN85oh  0.1  KK6DJC  DM12kw  1459  159 

73,

Steve

KF7O

[ZL2APV]

Hi Steve,

I see you copied my whisper signal ...

Timestamp	Call	MHz	SNR	Drift	Grid	Pwr	Reporter	RGrid	km	az
2016-05-01 07:10	ZL2APV	14.097072	-25	0	RF70bw	5	KF7O	CN85oh	11296	40

Interesting that there was no drift on Rx for you so it really is only the heat build up on Tx that is the problem. For my part I am not worried by 2 or 3 Hz in a Whisper 2 minute time frame. I have been using FreeDV but on my HiQSDR not the Hermes-Lite. It will be interesting to see if Tx drift will be a factor in that mode but I need to get my power out up a bit more to get into VK where the only other FreeDV station I have heard is located.

73, Graeme ZL2APV

[Steve Haynal]

Hi Graeme,

Great to see the WSPR spot! The frequency drift only occurs when switching between TX and RX as both use different amount of currents and heat the Si510 differently. I'll see drift after TX on the next RX cycle. I won't see drift on the second TX cycle immediately after a first TX cycle which will have drift.

73,

Steve

KF7O

[pascal.v...@gmail.com]

Hi John,
Just a small correction in the Digikey reference in the BOM  = FB2    Ferrite chip, MPZ2012S221A    445-1568-1-ND

[pascal.v...@gmail.com]

I will order the parts of the 1.42 BOM from Digikey next week, if any european builder is interested, I can do to share the shipping cost - Let me know by a private email

[John Williams]

thanks for the update, Pascal. Revised CSV attached...

John

On 5/2/2016 7:05 AM, pascal.v...@gmail.com wrote:

Hi John,
Just a small correction in the Digikey reference in the BOM  = FB2    Ferrite chip, MPZ2012S221A    445-1568-1-ND

[Serge K]

Helo Jim 

I have now   PCB 1.32 Front end    please where I can  find   BOM  for 1.32 ??

Thanks

73

Serge

четверг, 21 апреля 2016 г., 19:06:37 UTC+3 пользователь James Ahlstrom написал:

[John Williams]

Serge,

I would suggest you await the 1.42 board. Much better results. That is what I will be doing.

Board will be sent in the post today.

Here is the bom anyway.

John - W9JSW

--

[Serge K]

John

I agree

keep me  updated and reserve two PCB board   v1.42

73! Serge RV3APM

04.05.2016 13:43, John Williams пишет:

You received this message because you are subscribed to a topic in the Google Groups "Hermes-Lite" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/hermes-lite/fU1DCAAjOYg/unsubscribe.
To unsubscribe from this group and all its topics, send an email to hermes-lite...@googlegroups.com.

[pascal.v...@gmail.com]

Hi John,

PCB received , thanks,
On the BOM this digikey ref for the SMA is not the right one

2,rfconnectorB,SMA/BNC/HDR,P2 P3,SMA RF connector,WM2744-ND

[ZL2APV]

Hi John,

Got my board etc. today. Many thanks for this John - a very generous service.

73, Graeme ZL2APV

[Glenn P]

also received kit, many thanks John

glenn

[ZL2APV]

C16 is shown as 150 pF on the BOM and 180 pF on the circuit diagram. Which one is correct please?

Graeme
ZL2APV

On Tuesday, April 5, 2016 at 5:51:30 AM UTC+12, James Ahlstrom wrote:

Hello Group,

I would like to order 1.41 boards within the next few days.  It takes some time for them to come.  We can still change component values.  Does anyone have changes other than component values?

Jim

N2ADR

[Sid Boyce]

Same here.
Thanks.
73 ... Sid.

> --
> 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]

BOM also says the op-amp is a SN10501 while , Schematic and parts supplied are OPA2674.

The schematic and PCB overlays need to be tidied up also and re-issued I think. A lot of the text is obliterated on both.

glenn
vk3pe

On Thursday, May 12, 2016 at 9:41:51 AM UTC+10, ZL2APV wrote:

C16 is shown as 150 pF on the BOM and 180 pF on the circuit diagram. Which one is correct please?

Graeme
ZL2APV

R

[James Ahlstrom]

I am away until May 16, so I don't have my notes. But I believe the schematic is correct as 180p. This is Andrew's filter as published on this group. Please follow the schematic and look at frontend.html in the directory.

Jim
N2ADR