Home brew T/R relay switch for Anan 10 pure signal

[W3RSW]

I've built a T/R relay board based on the TAPR TR Plus but having just one larger Shrack relay instead of the two smaller ones in parallel that TAPR uses.

http://www.tapr.org/kits_trplus.html

A 43 dB sampler with additional 10 dB attenuator on the sample output line is connected to the Alpha thence back to the Anan RX input (Hermes board direct) via the T/R switch.

Even though I just trial-lashed it up without a shielding box it Works surprisingly well; hung it off the back of the Anan with bracket and existing panel screws.  12 watts pep input to my Alpha 89 yields 800 out.  Tha Anan automatically cuts carrier to one fourth PEP in AM service and cleans up not only IM products but also AF overdrive, if any, in modulation (within limits of course  . )

I'll put up some panafall screen shots later.

[KD6VXI]

Very interested.   I was going to do a single bottle 800A7 until I scored a trifecta of 8930s.  So,  I'll be doing an anan 10 with tetrode,  come tax time.

--Shane
KD6VXI

[W3RSW]

Nice tube but be careful with the screen as you probably already know.
See the notes.

http://www.g8wrb.org/data/Eimac/8930.pdf

You could do worse than getting a GM3SEK tetrode control board. Does everything and utilizes SS relays for inrush protecting of HV /screen supply, main power relay, screen protection, etc.

I retrofitted my 8875 HB pair amp with his triode control board and am very pleased.
If you already have a 3cx800A7, I'd go with that and the triode control board.  Way simpler circuit and good match with Anan 10 for 500 to 750 pep out, about same as  what you'll get from a pair of 8930's.

Here's one built almost back in the day with grid protection. Overbuilt for control symmetry with chain / sprocket drive, etc. , but a good place to start. Easy 600 pep with an Anan 10 .
If building today I'd get PI net inputs from relay board circuit.
http://www.robkalmeijer.nl/techniek/electronica/radiotechniek/hambladen/qst/1988/07/page22/index.html

[KD6VXI]

Rick,

I had planned on substituting a 3cpx800 into this 4cx250B based Pride.   The fil and bias xformer zorched and with a seperate 2.4kv DC supply it was a no brainer.

Then the price of 800s went out the window.   So I had planned on something else.   Then the three 8930s came up on this board.  Since there are a couple 4cx250 based amps laying around,  made sense.

This will be the driver.   I have plenty to make legal limit plus,  including a 3cx3k.   Need 3 to 400 for drive.   A single 8930 should rock at that.   When set up for the 8877, I expect mucho cleanliness!   An 8930, delivering 65 watts pep with pure distortion and an 8877, should be about as clean as Tom's class A lash up.

--Shane
KD6VXI

[W1AEX]

Nice and clean Rick and I like the heavier single relay rather than using the two smaller parallel connected relays. It certainly looks more compact and much neater than my homebrew junkbox PS relay solution! I ended up with the same -43 dB plus -10 dB attenuation pad arrangement and have found it makes the whole setup much happier than -43 dB + 20 dB as it presents an RF sample with more headroom above any of the rig's internal noise that can be a factor on the upper bands. What kind of 3rd order IMD levels are you seeing with pre-distortion engaged?

Yesterday, I was messing around with connecting my SDRplay to the second port of my AC2IQ coupler (with an additional 20 dB attenuation) to see if I could do some video captures of the ANAN + OpenHPSDR correction on sideband with voice rather than the usual two-tone test. I was seeing signal levels exceeding 50 dB above the spectrum baseline with no signs of blooming on either side of the signal. I generally don't run pre-distortion on AM simply because it runs that mode very cleanly without it but you are right about the way it removes all distortion, even at the audio level. Warren (NR0V) indicated that THD drops from a nominal 3% to around .2% when pre-distortion is engaged.

73,

Rob W1AEX

[KD6VXI]

Rob,

Are those numbers out of the Anan itself,  or with an amp after?

Trying to see what I should expect.

--Shane
KD6VXI

[W1AEX]

Shane, that is what is seen with pre-distortion wrapped around an external amplifier. In the screenshot above I was running 600 watts PEP with an ALS-600. With a good RF sample from the output of an external amplifier the PS correction does a great job of correcting the entire transmitter chain. It varies from band-to-band but with a two-tone test I can reliably get 3rd order down -50 dB or better with the ALS-600 and with the AL-82 at full legal limit the numbers are slightly better. The latest algorithm does a beautiful job of correction with voice modes and handles amplitude changes on-the-fly very reliably.

The video that I pulled the screenshot from can be seen at the link below. The first part shows SSB with pre-distortion enabled and the second part shows AM without any pre-distortion.

Rob W1AEX

https://www.youtube.com/watch?v=H95GKf6o_r4

[AB2EZ]

Attached is an analysis I did on the effect of crosstalk from RF signals from sources prior to the output amplifier that add, at the input to the ANAN receiver, to the sampled output of the amplifier.

The bottom line:

If you want to improve the IMD, using Pure Signal, by X dB, compared to what the amplifier's output IMD is without Pure Signal ...

Then, it necessary that the feedback signal being injected into the receiver from the amplifier's output sampler be stronger than the crosstalk (RF from sources before the amplifier's output) being injected into the receiver by at least X dB

Remember, the ANAN's display corresponds to the total signal being injected into its receiver (amplifier sample + crosstalk)... and the fact that the display shows very low IMD products (associated with the total signal being injected into the ANAN's receiver) does not ensure that the amplifier's output IMD is also that low.

Stu

[AB2EZ]

As an example of the issue I raised in my prior post:

If you want Pure Signal to be able to reduce the IMD at the output of the final amplifier by 20dB (compared to what the IMD is without Pure Signal)... then the sample of the output being applied to the input of the ANAN's receiver has to be at least 20dB stronger than the crosstalk being coupled into the ANAN's receiver from the transmit side of the T/R relay to the receive side of the T/R relay.

If the gain of the external amplifier is 18dB (12W in => 800W out), and the coupler has 43dB of coupling loss, and the added series attenuation is 10dB... then the sample of the output being fed back to the ANAN's receiver is (43+10-18) dB = 35dB below the RF signal level on the transmit side of the T/R relay.

To achieve the necessary 20dB of contrast between the output sample and the crosstalk, the T/R relay isolation has to be at least 55dB.

Stu

[W3RSW]

Thanks for the attenuation in series write-up Stu.  Very erudite.
I've also taken the 10 dB add'l attenuation out of line using only the 43 dB of the sampler when running 12 in, 800 out with the Alpha 89.
Still get 'green' within range of auto sampler in pure signal feedback indicator although dynamic range is different.  And yes, without shielding of the HB T/R board, I'm probably getting 'wrong' RF feedback more than optimum.  And don't forget an add'l 20 dB Hermes switchable Rx input attenuation,( although it's really the preamp always on with the -20 dB switchable.)

Regardless of perfection and the originating stations, ALL the signals I've seen on panafall are always cleaner, straighter edged, (some almost perfectly in the waterfall) when pure signal has been applied to an Anan 100 or Anan/Amplifier combo.  This includes the tribe that doesn't use an outboard T/R device like mine, but simply use the Anan's internal pickup of about three inches of coax going from the Rx (Hermes) input to the grounded on transmit internal relay.  They add 50 ohms in series to the Rx port and get feedback from 50 -5jx roughly of the internal link.  Virtually every signal seen with pure signal must be getting at least 20 to 30 dB improvement of IM3, 5, etc.   Interestingly some further out IM's, say 9, 11, etc. do not get much improvement if any, and some seem slightly worse.  That's because pure signal only works over, say a 20 kHz range.

Now when I tried to use pure signal with the base Anan 10, no following amplifier, with my only sampler of -43dB, no combination of internal /external attenuation would get sufficient feedback to get the auto
ranger out of the "red."  Nice try Rick but I tried even knowing it wouldn't be sufficient, but it's a dynamic system so some reduction of IM 3, 5 was noticed left over from previous constants I suppose.
For those not yet using pure sampling, constants can be saved in a file for next time usage although the auto-ranging is so quick it doesn't seem to be worthwhile to call back old stuff from a file.

And as far as phase goes, here's a magnified view of what the Anan 10 / Alpha 89 combo produces as well as linearity correction (which obviously was pretty good to start with.)

[VE3ELQ]

Rick that looks really good.  One question though.  Was that a 2 tone test or voice modulation.  With my 10E and HB amp  the 2 tone looks similar but not quite as good as yours but the voice modulation makes a squished figure eight.
73s  Nigel

[AB2EZ]

Rick

Please note:

The panadaptor and panafall displays indicate the IMD products at the input to the ANAN's receiver... those displays do not indicate the IMD products at the output of the external amplifier unless the sample of the amplifier output being fed into the ANAN's receiver is much larger than the crosstalk being fed into the ANAN's receiver from sources like the T/R relay that have not passed through the amplifier.

The ANAN's internal attenuator attenuates both the sample arriving from the amplifier and the crosstalk... by the same amount. Therefore that attenuator does not affect the contrast (dB difference) between the sample of the output and the crosstalk.

Stu

[AB2EZ]

Rick

Note that the AmpView screen capture that you posted looks suspiciously good. It is surprising that the required correction of the gain is so small when the amplifier is producing 800W of peak output.

Separately...

Since your amplifier has a gain of approximately 18dB (12W=>800W), you might want to do the following experiment:

a. With the amplifier in line, and Pure Signal active (auto-attenuate active)... note the value of the S-ATT when transmitting (2-tone test, click on "single calibrate"). If the sample being fed back from the amplifier to the input of the receiver is sufficient... the S-ATT value should be at least 18dB (but, to avoid damaging the amplifier... not more than 30dB)

b. Then bypass the amplifier, but leave everything else the same. I.e. the -43dB coupler, the T/R switch, etc. are still in place. The amplifier has been effectively replaced by a straight piece of coaxial cable, but that is accomplished by the amplifier's internal "bypass" relay.

Note the value of the S-ATT when transmitting (2-tone test, click on "single calibrate").

The S-ATT value (when transmitting) should be 18dB lower when the amplifier is bypassed... corresponding to the removal of the 18dB of amplifier gain from the feedback path.

[In my case, with my ANAN-10E and my Elecraft KXPA100 amplifier... the values I see are: S-ATT = 17dB and S-ATT = 4dB... corresponding to the 13dB gain of the amplifier]

As another useful test:

c. With the amplifier in line, and Pure Signal not active (off)  and with "DUP" active: note the value of the reading of the (single carrier) signal on the Panadapter when transmitting ("tune"). This should be in the range of between 0dBm and +13dBm. [Much more indicates potential damage to the internal attenuator at the input to the ANAN's receiver]

d. Now disconnect the cable that is coming from the -43dB coupler, but leave the other end connected to the receive side of the T/R relay. Terminate the disconnected end of this cable with a 50 ohm termination. With the amplifier bypassed (off), and Pure Signal off, and with "DUP" active, note the value of the reading of the (single carrier) signal on the Panadapter when transmitting ("tune").

The reading observed in d. (i.e. just crosstalk, no sample from the coupler's -43dB port), should be at least X dB lower than the reading observed in c. if you want Pure Signal to be able to produce X dB of IMD improvement.

[In my case, I see +5dBm with the amplifier on and the output sample connected to the ANAN's receiver input. I see -25dBm with the coaxial cable connection between the output sample and the ANAN's receive input removed (with or without the 100W amplifier in line). Therefore I am seeing about 30dB of contrast, and I can expect up to 30dB of linearity improvement using Pure Signal. My 40m AmpView screen capture is attached. Without Pure Signal, and on 40m, the KXPA100 3rd order IMD products are only about 20dB below the principal tones with a 2-tone test (100W peak output). With Pure Signal, the display shows the 3rd order IMD products down by more than 50dB. Since I have measured 30dB of contrast between the amplifier output feedback signal and the crosstalk... I believe that the Panadapter display of the IMD is a good representation of the actual IMD at the output of the amplifier.]  

Stu

P.S. For every complex question, there is a simple answer... that is wrong.

[W3RSW]

Hey Stu, I'll carefully read your comments and analysis. I've seen others adding the 20 dB Hermes attenuation in the series expansion so I mentioned that.  What you have shown makes sense and I'll study it further.  I'll also try your suggested experiments.  And yes the excellent graph I obtained was for the compensated RF being fed back, not the actual signal going out on the air.

That's most of the reason that I added the remark about ALL the pure signals seen on the air as being better than regular. It is simply that pure signal apparently doesn't have to be perfect in order to have some action. The auto calibrate adjusts what it can on the fly with what signal it can mathematically muster and even though showing "red" still tries to work.
It will also show "green" or go with better but still not optimum signals because the auto calibrate function uses what it can

I was amazed at how easy it is to get up and running even though it looks complicated. Once the sampler is in line, the T/R logic wired up for your chosen system, whether it be direct to the Hermes RX port for the Anan 100, or with external T/R for a lesser output xmitter such as the Anan 10, it just works simply by calling it up and pressing a few buttons. I wanted people not to think of it a excessively complicated nor married to an absolute 55 dB differential.  It simply works at some level of competence for all who've used it.

And As you know, the 3cx800a7s at only 12 watts input, six watts per tube are already pretty linear devices, straight lines on Eimac's curves, load line, etc.  that coupled with rock solid 2450 VDC supply at Eimac's original , "pre FCC laden" specs., with such light loading of the power HV supply
http://www.g8wrb.org/data/Eimac/3CX800A7-Proposed-FCC-Ruling-May-1985.pdf

And for curves, extend the 0.01 grid line: (pun)
http://www.g8wrb.org/data/Eimac/3CX800A7.pdf

[W3RSW]

Nigel, it was two tone.

-now printing out to study closer Stu's notes.

[AB2EZ]

Rick

Okay.

I can see that the Alpha 89 (two tubes, in parallel... grounded grids) would be very linear at 400W output per tube.

Eimac specifies the 3rd order IMD products at 36dB below each principal tone in a 2-tone test, with 750W output (1 tube)... so with 400W/tube the 3rd order IMD performance could be as much as 5.5dB better. (I.e. with a simple model of the non-linearity: the 3rd order products would decrease 3dB for every 1dB decrease in the principal tones, in a 2-tone test).

With my Elecraft KXPA100 amplifier bypassed, and the ANAN-10E putting out 5W peak power... and with Pure Signal turned off... the 3rd order IMD products are 40dB below the principal tones in a 2-tone test. This would be similar to what you have with 400W per tube (i.e. 3rd order products down about 36dB + 5.5dB = 41.5dB from the principal tones, without Pure Signal).

When I turn on Pure Signal, the 3rd order products (as displayed on the Panadapter) go down an additional 25dB. But, given the 17dB of contrast between the ANAN-10's internal crosstalk and the sample of the RF output being produced by the output coupler that I am using (when the Elecraft amplifier is bypassed), I expect that the real improvement in the output IMD might be somewhat less than 25dB... depending on the details of where, inside the ANAN-10E, the crosstalk is coming from. I.e. is the crosstalk all coming from the output of the ANAN's internal power amplifier, or is some of it coming from the input to the ANAN's internal power amplifier.

The AmpView display is essentially flat (see attached).

It will be interesting to hear the results of the experiments/tests I suggested.

Stu

[W3RSW]

Stu,
There is no variable "S-ATT" steps settable in the Anan 10 as opposed to an Anan 100.  "ATT" button only is either 0 or 20dB.
Where are you reading variable "S-ATT" in the 10e?  From the saved at op. conditions file if you chose to save?

[AB2EZ]

Rick

If you move your mouse pointer to be over the letters: S-ATT (the letters, not the box underneath the letters)... and if you double click... the S-ATT will switch from 10dB steps to 1dB steps.

When you open the Linearity (Pure Signal) window, you will see a box labeled "Auto Attenuate". Make sure that box is checked.  if you click on "Two-Tone Gen" and then "Single Calibrate", the software will automatically adjust the "S-ATT" (receiver's input attenuator) to the correct value, and the associated reading (on transmit, npt receive) to provide the correct signal level to the receiver when you transmit. I.e. the S-ATT reading will change when you switch from receive to transmit. After that, whenever you transmit, the S-ATT (setting and reading) on transmit (not on receive) will revert to that Auto-Attenuate value. If you change the transmit power level, you need to, again,  click on "Two-Tone Gen" and "Single Calbibrate" to readjust the S-ATT setting to accomodate  the new value of received power being fed into the receiver by the sampler/coupler.

Stu

[W3RSW]

ok, I did see that before.
and all the thread comments before and after.
https://groups.yahoo.com/neo/groups/apache-labs/conversations/topics/25764
 It's the resistance step attenuator. Just forgot the double click trick.  Also the description changes automatically from ATT to S-ATT on transmit.
Anyway I'm getting only an action diff. of 14 dB.  Apparently all the unshielded external plus internal RF diminishes action. It's Pretty interesting. The combo's pretty clean without pure signal, but now somewhat better. Really appreciate all your helpful guidance along with yours and Scott's, etc. notes.

Oh again reading those notes, my 43 dB sampler may be too much attenuation for what Scott calls the little people amplifiers.  What are you using for a sampler off the KPA 100 back to the 10e?  I get auto range green with the 10 dB add'l in line but blinking blue (over range) without it, after setting the initial fixed RX 20dB attn to 0 dB.  When I leave the 20bB in RX line, both cases are green in auto.
 

[AB2EZ]

Rick

In your latest post, you have referred to an "action diff." of 14dB. Please define what you mean by an "action diff".

You also refer to an "initial fixed attenuation" of 20dB. Please note that the ANAN-10 (and the ANAN-10E) does not have both a fixed attenuator with 20dB steps and an adjustable attenuator with 1dB steps. There is only one attenuator...which can be set to either 20dB steps (ATT = 0dB or ATT= -20dB) or 1dB steps (S-ATT = 1,2,...,31) in receive mode. The attenuator setting you choose to use in receive mode has no effect on the S-ATT setting that Pure Signal selects, when transmitting, if "Auto Attenuate" is active, and has no effect on the behavior of Pure Signal, or the Panadapter reading (when transmitting).

In the case of my ANAN-10E, I am using at total of 45dB of attenuation between the 100W PEP output of my KXPA100, and the RX input of the ANAN-10E. Therefore, the sample being injected into the ANAN-10E is +5dBm = 3mW PEP.

Stu

[W3RSW]

Stu, I see your using KC9XG drawing or equivalent sampling circuit.
https://groups.yahoo.com/neo/groups/apache-labs/conversations/messages/25774

In that message you've stated that his sampling was insufficient.

My setup is very similar. The -14 dB I mentioned came from my 800 watt post amp.
 59Dbm -(43+10+20) = -14, ( where 43 sampler, 10 add'l attn, 20 internal relay grounded.)
Auto calibrate likes it very well ( green) as it is only one dB above the ADC's -15 to -18 favorable range and below the ADC's -12dBm overload range.

Proving that, when I remove the add'l 10 dB commercial in line attenuator, the sampling indicator shows blinking blue.  All this when setting the two tone at 0.00 dB.
Not sure what more I can do to satisfy pure signal requirements, given the sampling indicator being so favorable.
Those on air say it's a straight line edge stop to band pass.

[AB2EZ]

Under the conditions in your last post, with the 10dB series attenuator in place, and the Pure Signal level light showing a green indication... what is the S-ATT reading when transmitting with the 2-tone test?

Note:

If you have removed the internal mini-coax jumper that connects K19 to the input of RX1 (specifically to the input of the step attenuator), then you will not have the additional 20dB (on 40m) of attenuation of the sample being fed back from the coupler when K19 is in the transmit position. [The 20dB of attenuation is associated with the voltage divider formed at the RX1 input by the inductive reactance, looking into the mini cable, when the K19 end is shorted by K19 on transmit. If the mini cable has been removed, this effect does not occur]

If you have not removed that jumper, then there is no need for the external T/R relay that you are using. I.e. the external T/R relay would just duplicate the the T/R functionality that is already present. In that case, a direct connection between the output of the coupler, with the extra 10dB attenuator in series, and the RX input, would work just as well. The only effect of not using the external T/R relay would be the add a negligible amount of thermal noise to the input of the receiver, when receiving. If have the mini coaxial jumper in place (between K19 and the input to the receiver), then you should add an external 50 ohm non inductive resistor in series with (not across) the RX input to the ANAN-10. This resistor should be located close to the RX input connector. I.e. keep the length of the coaxial connection between the RX SMA connector and the 50 ohm series resistor to 2 inches or less.



Stu

[W3RSW]

That's right. Nothing's been done to tha Anan 10 internally. I have +6dB pure signal feedback going into the Rear RX port.

And yes a direct feed with 50 ohm (50 -5jx)  is a good alternative. I built the board incase of internal relay failure.  ( full 6dBm into the ADC.).  
From other posts I see that the 50 ohm should be non-inductive and your other notes.

[AB2EZ]

Rick

Okay...

Stu

[AB2EZ]

Okay...

If I understand your last post, you have added the internal T/R switch as a precaution, in case K19 fails to short the K-19 end of the mini-cable on transmit. However, at this point, you believe that K19 is working as it should. [Note: I don't see why the external T/R relay would provide any defense against this K19 failure mode. When you are transmitting, the feedback path from the coupler to the RX input is the same... with or without the external T/R relay]

If so, then something about the numbers you have posted doesn't "add up".

If you have:

+6dBm, nominal PEP output, from the 10dB external series attenuator (if it is looking into a 50 ohm load)

And, if you have

20dB of attenuation caused by the voltage divider formed at the input to RX1 by the presence of the shorted (at the K19 end) mini-coaxial cable

Then you have:

 -14dBm at the input to the RX1 step attenuator.*

[* I find it confusing when power is inadvertently expressed in dB instead of dBm]

That implies that if "Auto Attenuate" is working, the S-ATT reading during a 2-tone transmit test, after having clicked on "Single Calibrate", would be around 3 (i.e. 3dB of additional attenuation from the step attenuator, when transmitting). The "Auto Attenuate" process is trying to adjust the step attenuator so that the input level to the A/D converter is approximately -17dBm... i.e. so that the output of the step attenuator is -17dBm.

An S-ATT reading of 3dB implies too low a level of feedback from the output of the amplifier. I.e. the feedback level being fed to the input of the A/D converter is okay (around -17dBm PEP)... but the contrast is too low (too small a difference in dB) between: the feedback being fed to the receiver from the output of the amplifier (via the coupler) and the crosstalk being fed to the receiver from sources prior to the final amplifier's input.

If the S-ATT reading (during a 2-tone transmit test, after having clicked on "Single Calibrate") is significantly higher than 3dB (example: around 15dB-20dB), then the feedback level is okay... but the actual attenuation in the feedback path is less than you think it is. For example, perhaps K19 is not working, as it should, to short the K19 end of the coaxial mini cable on transmit. As another example, perhaps you are operating on a higher frequency than 7MHz. E.g., on 14MHz, the effect of the voltage divider formed by the shorted (at one end) coaxial mini cable would be to add only 20dB-6dB= 14dB of attenuation... so the feedback path would have 6dB less attenuation.

Separately, if you have checked Auto Attenuate, and if you have activated the 2-tone test, and if you have clicked on "Single Calibrate"... then the S-ATT should be able to add enough additional attenuation to compensate for the removal of the extra 10dB series attenuator. I.e. the Pure Signal "Feedback Level" light would only be blue for a second or two. Then (if you have activated the 2-tone test, and if you have clicked on "Single Calibrate") it would turn green as the step attenuator adds in 10dB of additional attenuation.

Stu