Retro 75 Modulation Frequency Response

[AB2EZ]

Hi!

I have talked, on several recent occasions with Chuck, W3FJJ... when he was using his new Retro 75. It sounds very nice.

I was looking at the schematic of the transmitter, and I found two places that constrain the low frequency response.

The first is C33... the 100 uF capacitor that provides DC blocking between the output of the audio amplifier and the input to the modulation transformer.

According to the Retro 75 manual, this transformer is nominally 3.2 Ohms:48 Ohms... so the turns ratio is approximately 1:3.9.

If I assume that the B+ on the IRF530A is 12 volts, and that the average current is 0.3A at carrier (i.e. 3.6 watts DC input and ~2 watts rf output)... then the modulation resistance of the rf stage is about 40 Ohms. Thus, the impedance, looking into the modulation transformer is approximately 40 x 3.2/48 Ohms ~ 2.7 Ohms.

The question is: at what frequency is the impedance of C33 (100uF) equal to 2.7 Ohms. The answer is 589 Hz.

Even if the impedance looking into the modulaton transformer is somewhat higher, the low frequency cutoff is at too high a frequency.

I would suggest increasing C33 to 1000uF (or at least 470 uF)

The second low frequency bottleneck is C24. The gain of the audio preamplifier is 10 at medium frequencies (e.g. at 500Hz), and 20 at frequencies above 1kHz; but, at the frequency where the impedance of C24 increases to 10k Ohms, the gain is 3dB lower than the medium frequency gain... and rolls off, as the frequency gets lower, at 6dB per octave.

The frequency at which C24 has an impedance of 10k Ohms is 159 Hz. I would suggest increasing C24 from 0.1uF to 0.33uF

Stu

[WD5JKO]

Stu,

 With all do respect to your analysis, I'm sure the low frequency pole by design has something to do about increasing intelligibility to give a Retro-75 a fighting chance to make a contact with only 2 watts AM. I would hesitate to embrace East Coast Audio principles when the carrier is only 2 watts.

  On a side note, the power level of a Retro75 is a perfect match for a Central Electronics 600L single 813 broad band linear amplifier. That 813 will do 100 watts AM linear mode with a little blushing. Once a Retro75 is up to 100 watts or so, then increasing the low frequency response might be more desirable.

Just my opinion,
Jim
WD5JKO

[Superhet66]

Thanks
I've been saving my nickels for a retro-75. I've avoided temptation and passed on other QRP kits out there.

I have yet to hear one on air but I like the idea of increased low freq resp.

I recently journeyed down to middle NJ and picked up a 1957 handbook based 813 transmitter / modulator rack project. It was on Craigslist and no one was biting.
I'll post picks if anyone is interested.
It might be interesting to bumb it with the Retro-75.

re: the 600L, It does seem like a match made in heaven.
( in a non-denominational way   )

http://www.eham.net/reviews/detail/3872

[W1RKW]

Stu,
I'll have to give those mods a try.
Bob

[kg8lb]

Using mine with a D-104 feeding a K7DYY mic processor. I switch the same mic/amp to all of my transmitters. In the case of the Retro 75 I had to install an inline attenuator as the K7DYY output was a bit too high. I also installed a micro reed relay for PTT function as my station's switching system directs both PTT keying and the mic audio.

 Really enjoying the little bugger and aside from the PTT I have no present intention of any further mods. The concept is quite sound and works well. I have no inclination to encroach on the elegance of the design by trying to make it sound like a 2 watt broadcast transmitter. This is indeed a real piece of amateur radio gear.

[WD5JKO]

The first is C33... the 100 uF capacitor that provides DC blocking between the output of the audio amplifier and the input to the modulation transformer.

   Stu, I retract what I said earlier in this thread. I did some testing of my Retro75 last night, and it becomes apparent that the whole speech amp including the recommended microphone are designed to work together.

    The C33 issue is interesting. That little tiny 100uf cap besides being low in value, also likely has high ESR. If you disable the automatic modulation control, and increase C33, a low frequency instability occurs, even when driving with speech. The problem appears to focus around the modulation transformer which seems to saturate is the frequency drops below 500 hz. To me this is Deja Vu back to the Gonset G50 where the Heising choke saturates below 500 hz.

    If I just drive the modulator by itself (no preamp), I still see the effect. Increasing C33 also requires increasing the +12V decoupling capacitor to > 1000uf, and when doing that things get better noticeably. With C33 at 1000 uf, C112 bridged with 1000uf, and C110 bridged with 330 uf, the modulator is flat 500-5000hz and will very cleanly modulate 100%. Going below 500 hz you see the waveform distort as the % mod lessens, and the power supply current increase, i.e. core saturation.

   So the trick is to Taylor the speech amp to lessen the energy below 500 hz to stay away from saturation. After the mods, I had the following behavior with a nice modulated envelope pattern (driving U6 directly from empty socket at U5-1):

150hz, 25%
200hz 40%
300hz, 70%
500hz - 5000hz, 100%
5000hz - 8000hz, 90%

   In stock form the modulation limiter prevented a lot of the bad behavior by limiting the modulation percentage. In my case with a sine wave drive, I am limited to 30% modulation with an asymmetrical waveform (distorted). Speech will hit higher peaks though.

   I need to take a fresh look at the preamp stage. With the changes I made to the modulator (U6), I see instability (low frequency)* if I run audio as intended into U5. I think there could be a ground loop or some other form of regenerative feedback. Still all the bad behavior comes about below 500 hz.

   It might be easier to eliminate the preamp, and use line level input directly to U6 with a gain pot added. I have an external speech amp ready for that task.

   Been thinking of modified Heising to get the DC current out of the mod transformer.

* exceed a certain level at < 500 hz, and the waveform increases in value suddenly with asymmetrical distortion. Bring C33 back to 100 uf and all that weird stuff goes away.

This is a very interesting little rig. I am amazed at Small wonder Labs that they are producing this little rig for such an inexpensive price. A lot of thought and work went into it. I can see a lot of fun ahead with this nifty little rig.

The mod transformer used is a Xicon 42TU04C-RC, rated 48/8 ohms at 460mw with a low end down 3db at 300hz. A possible alternative is a Hammond 106R is also 48/8 but at 1500mw with a low end down 1 db at 170hz.

http://www.mouser.com/catalog/specsheets/XC-600133.pdf
http://www.hammondmfg.com/102.htm


Jim
WD5JKO  

[AB2EZ]

Jim

Also reminiscent of the Johnson Ranger....

When you remove some of the high frequency bottlenecks in the audio chain and the modulator... you discover that some of those bottlenecks are also "band aids" that the engineers used to control oscillations associated with nonlinearities in the high level portions of the audio chain + the audio feedback circuit. Then you have to fix those oscillation problems (a different way) as well.

Stu

[Steve - K4HX]

Usually the cause is/was that the decoupling caps between the audio stages were too small. People often forget to change those when doing audio mods.

[Bill, KD0HG]

A Johnson T'Bolt linear would also be a perfect companion. A grid driven pair of 4-400s. Find a CB op willing to part with one.

[WD5JKO]

Some Retro75 updates.

   Since last week I finally got back out in the shack since the temp was in the 60's until a little while ago. Boy it was cold just a few days ago.

   Anyway, I made a few discoveries.

That Hammond transformer I linked to costs ~ $40 whereas the Xicon that comes stock sells for about $2. Big difference! Looking at the schematic I thought it would be easy to disconnect the mod transformer primary from C33 (-), and then connect C33 (-) to the secondary C.T.  This was easy. The idea was to use the secondary as a tapped Heising reactor. The results were interesting.

Stock 2.45:1 mod tranny:
The stock tranny used as intended is 48-8 ohms, or 48/8, square root of 6 = 2.45 turns ratio. So about 5 v peak from the modulator will provide ~ 12v peak to the RF amplifier to 100% modulate it. The modulator with 12V on it, idles at 6V, and is capable of almost +/- 6V to swing 12v P-P.

Same tranny used as a Heising reactor:
Since I am driving the audio to the C.T., I expected the turns ratio to be 2:1 which would lower the maximum modulation percentage to ~ 82% if I had 100% before. That is pretty much what happened but with a big surprise! The waveform distortion below 300Hz limiting the low frequency max mod % got better, and by a lot. Also the modulator current was lower than before, and the low frequency instability was much better, but still could be coaxed to occur, but at a lower freq (200hz) and input higher level than before.

Searching for the low frequency instability when using the mic preamp turned out to be easy. At Pin3 of U5 we pre-bias the (+) input to 6V from the U2 regulator. Chopping that trace (near TP1) and inserting a 4.7K resistor, along with replacing C108 (0.1uf) with a 10Uf cap seemed to completely eliminate the low frequency instability. I later found it to return somewhat when putting the mod tranny configuration back to stock, but still much better than before adding the R-C.

So I did some audio tests driving U6 only as I did before. With the mod tranny as a C.T. Heising choke, I could modulate cleanly at 85% from below 300Hz to above 5 khz. At 1 khz, sine, triangle, and square wave envelope patters were darn good at 85%, and near perfect at 50%. Overdrive to 100% was possible with clipping. Looking at U6 output it was clipping at 6V peak both up and down as you'd expect. To modulate higher without clipping, the turns ratio needs to be higher. That would mean moving the C.T. away from center.

When driving the mic input through U5, I did some tests. Remember my last post (driving U6 only) where C33 is 1000uf, and I bridged 1000uf across C112, and replaced C110 with 330 uf.

Max undistorted mod % when driving the mod tranny at the C.T. (Heising):

100hz - 20%
150 - 50
200 - 80
300 - 85
500 - 85
1000 - 85
3000 - 85
5000 - 80

As above, but Mod tranny back to stock:

100hz - 10%*
150 - 20*
200 - 50*
300 - 85
500 - 100
1000 - 100
3000 - 100
5000 - 90

* increasing level from here causes LF instability where every odd cycle peak has more amplitude than the even cycle peaks with a ratio of almost 2:1.

  I tried three microphones (2 dynamic and D104). The audio polarity seemed backwards with the dynamics, and monitoring myself (always subjective) sounded pretty raspy to me since the audio peaks were easily clipping the baseline. Reconnecting the modulator AGC corrected that problem, and the audio sounded decent. The D104 output was low and tinny sounding. Using a crystal mic requires a higher load impedance, and with that a level control.

Edit: A frequency sweep from Mic input as modified was 200-4000 at -3db referenced to level at 1500 hz (peak). This was done at 50% modulation looking for the .707 level on the scope. The freq response was the same with either mod transformer configuration.


All in all a fun evening in Retro75 land.

Jim
WD5JKO

[steve_qix]

Hi Stu,

I've heard a few Retro 75s on the air, and immediately noticed the constrained low frequency response.  They may have done this "on purpose" (because it's low power), but I personally would have made the rig full fidelity (I also wouldn't have used a transformer, but that's another discussion!), and then either given an option to switch in a low cut filter, or something - as long as the rig was CAPABLE of high fidelity audio.

I'm wondering out loud if the modulation transformer can pass the low frequencies without saturating.  We will see 

Regards,

Steve

[AB2EZ]

Steve

Hi!

One nice things about these "little" transmitters... whether they are using solid state or vacuum tube devices... is that they are easy to tinker with... safe to tinker with... and we learn a lot by tinkering with them.

Even a "high end" audio transformer, operating at these power levels isn't going to cost very much ... and there is great satisfaction in tinkering with the design to achieve "broadcast quality" reports from a rig that was never intended to produce broadcast quality AM.

I must admit that I have occasionally "drawn the line" with respect to leaving a rig in its "as designed" configuration. I very much enjoy operating (once in a while) my Heathkit SB-10 sideband adapter in SSB or DSB with varying amounts of inserted carrier.

Stu

[WD5JKO]

I'm wondering out loud if the modulation transformer can pass the low frequencies without saturating.  We will see  

Steve, There is no doubt that the transformer is saturating below 500 Hz. This effect is countered with a pretty steep High Pass roll-off in the AC drive to the transformer, mainly C33 being 100uf. Then there seems to be a tendency towards a low frequency instability that is also held in check by modulator AGC loop.

That transformer is rated at 460 mw without any DC current flowing through the 48 ohm winding. We need a couple of watts of audio at least for 100% upward modulation. I have been scouring the internet looking for a suitable replacement, and short of a DIY wind job, all I see is the Hammond 106R that costs about $40 whereas the one used on the Retro75 (Xicon TU048-R) costs about $3 each.

So I have been moving forward trying to get the most out of the Xicon. In my last post I reconfigured the tranny as a tapped Heising reactor where the audio drives the C.T. That made a huge improvement in the LF response and linearity where it can be used down to ~ 200hz @ 80% modulation. The big drawback though was the turns ratio went from 2.45:1 to 2:1 which reduced the peak (midband) modulation from 100% to 85%.

So tonight drawing from a Timtron trick, I used the 8 ohm winding in series (phased properly) with the Heising reactor which should change the turns ratio from 2:1 to 2.8:1. Bingo!!
I can now modulate 120% upward midband with very clean waveform. Using voice with sharp audio peaks I see more.

In my last post I stated that the frequency response was 200 - 4000 at the -3db points with a deliberate rise to a peak at 1500 hz. Using the 8 ohm winding for more modulation didn't seem to change the response. Remember I stated earlier that C33 is now 1000 uf.

Next I need to control the negative modulation since that clips at the baseline very easily. While searching the 2, 3, and 4 diode negative peak limiters, I stumbled upon a circuit in my computer download file that I found somewhere. I will post it here. Anybody know who did this?

Edit: Added picture of 120% modulation at 1500 hz.

Edit2: added second picture with two diode clipper in action. I used a 1n5818 to handle the B+, and instead of 3 diodes to the keep alive supply, I am using a 68 ohm resistor in series with an LED. For the keep alive supply, I use U2 which provides 6v already.

Played some MP3's looking for that special male voice. Elvis Presley did the deed. No AGC action connected, peaks going to ~ 130% and occasionally more if overloaded. Listening on the receiver (Icom R75 wide open) it sounded pretty good considering that LED was blinking pretty regular. No shortage of lows..

So I am getting about 2.75W of carrier (power at 12.6V), and around 12W PEP.

Jim
WD5JKO

[kg8lb]

 I believe the supplied Xicon output transformer is listed at 48:3.2  ohms (475mW) . Regardless, that would make a Hammond 106Q (48:3.2 ohms @1.5 Watt)perhaps a better choice ?  Another possible , perhaps more suitable package is the Hammond 146B , a 48:3.2 ohm but rated at only 700mW. Even at the 700mW rating it is a lot beefier than the Xicon unit and the price is only $17.
 Either way, a 3.2 ohm primary should present a more efficient load for the 2003 audio IC output.

  

[W1RKW]

I ordered the $40 hammond xfmr to try. I'll try the 146 too.

[WD5JKO]

I believe the supplied Xicon output transformer is rated at 48:3.2  ohms (475mW) , not 48:8 ohms . That would make a Hammond 106Q (48:3.2 ohms @1.5 Watt)perhaps a better choice ?  Another possible , perhaps more suitable package is the Hammond 146B , a 48:3.2 ohm but rated at only 700mW. Even at the 700mW rating it is a lot beefier than the Xicon unit and the price is only $17.

  It does say 48/3.2 in the Retro instructions, but the Xicon model (link above in my earlier post) is mentioned in the parts list which is 48/8 where both windings have a C.T.,

 Take that 8 ohm C.T winding, and common to C.T. is 3.2 ohms. I did verify by voltage ratio that we do indeed have a 2.45 to 1 tuns ratio, which is same as 48/8 = 6, square root 6 = 2.45.

  I am very excited to hear how the Hammonds work out. Still for those of us on a budget (tight wads) reconfiguring the Xicon as a tapped Heising reactor seems to work pretty well and gives you a higher turns ratio (2.8:1) so that 120% upward modulation (midband and  up) is easily achieved.

  If you want strong full lows below 300 hz though, the Xicon won't cut the mustard. The possibility of using the Hammond as a tapped Heising reactor also exists.

  I hope to take a picture of how I rewire that Xicon, but that will have to wait a few days before I can get to it.

   Remember the other issues I've posted such as the +12v decoupling caps, C33, and the low frequency instability that popped up on mine as soon as I increased C33. All had solutions, and need to be addressed as a package, in my opinion.

Jim
WD5JKO

[kg8lb]

I guess my main point was that the previously mentioned alternative, the Hammond 106R was also an 8 ohm unit . I was suggesting that 3.2 ohm alternatives may be more suitable.(Notably the 106Q) One of the 3.2 ohm alternatives was quite a bit lower in cost as well.
  Of course the Xicon being an 8 ohm unit , using the CT could explain a better match as the modulator IC is very likely also a 4 ohm nominal output.(6 watts @ 4ohm, 10 watts @ 2 ohm loads)

[WD5JKO]

I guess my main point was that the previously mentioned alternaive, the Hammond 106R was also an 8 ohm unit . I was suggesting that 3.2 ohm alternatives may be more suitable.(Notably the 106Q) One of the 3.2 ohm alternatives was quite a bit lower in cost as well.
  Of course the Xicon being an 8 ohm unit , using the CT could explain a better match as the modulator IC is very likely also a 4 ohm nominal output.(6 watts @ 4ohm, 10 watts @ 2 ohm loads)

  Fun stuff! That 48/3.2 ohm tranny has a 3.9:1 turns ratio compared to the 2.45:1 in there now. The 2003 amp (U6) can drive a low Z as you say, but also realize it is a linear amp with almost no heatsink. Asking it to put out low volts at high current will mandate some thermal management since it might get spit boiling hot. :-)

  With the stock turns ratio, 5v peak drive gives you 12v peak to 100% modulate. Change the turns ratio to 3.9:1, and 3v peak gives you 12V peak on the secondary. Chip U6 can swing +/- 6V peak when B+ to box is ~ 13v. The 2 ohm resistance of the 48 ohm winding drops nearly 1 volt going to the FET.

Let me know how it goes, please.

EDIT: With a solid state AMP like this TDA2003, the source impedance is close to zero, so matching impedance's is not the route you need to go. The amplifier being class AB will achieve greatest efficiency when swinging as close to rail to rail as possible just prior to clipping. I prefer to use this idea to find the modulator turns ratio that gives me the peak modulated voltage I need, and at the same time swing rail to rail at the modulator side. If that were to come out to 2:1 as an example, then choose a ratio a little higher (maybe 2.2) to gain headroom. Keep in mind that the higher the ratio beyond what you need, the lower the modulator efficiency will be, and therefore it will get hotter. 

If you use the Hammond 48/3.2, the TR will be high, maybe too high unless you intend to modulate close to 200%. That 48/3.2 might be perfect though as a tapped Heising inductor phasing the 3.2 ohm winding like I am doing with the Xicon 48/8 ohm transformer.

Jim
WD5JKO

[W1RKW]

Someone please educate me on AM modulation and negative peak limiting, etc..  I was looking at Jim's scope images and see where he's getting 120% plus modulation out of his Retro75.  How does one determine modulation percentage beyond 100%?  And how does one determine it if the negative peak is clipped, soft clipped or not clipped?  The reason I ask is the images show the negative peak what I would consider clipped and almost soft clipped by the negative peak limiter he's experimenting with. To me it appears the negative peak flat-lined at baseline and that being the case, how does one determine that the envelope is actually at 120% in that condition?  Isn't there a subtraction occurring?  And if the baseline is clipped, what about the products produced?

I was messing around with a negative peak limiter myself but any differences I saw were very subtle and I wasn't seeing anything beyond what I considered 100% between non-modified Retro and modified Retro once the negative peak began to clip.  I was using just a signal generator and not a source that was asymmetrical.  Maybe that was my issue..

I would think if one is going to limit the negative peak, at baseline the sinusoid would be clamped and well rounded rather than what appears to be flat.

[k1swl]

Bob-

I'm not sure why it took me nearly a year to find this forum!  Interesting conversation going on. :)

100% modulation occurs when the minimum waveform amplitude just reaches zero. Assuming you know the no-modulation value and the modulation is symmetrical, the voltage swing from 'no-modulation' to 'positive peak' corresponds to 100%. Crank up the signal by 20% and you've got 120% modulation. The carrier then vanishes for a portion of the modulation waveform cycle.

I'll wonder aloud how much benefit comes from a keep-alive circuit at these low power levels.  Once overmodulation has set in, the audio 'envelope' is clearly distorted (both images). As such, the result is additional spurious content (i.e., splatter). I doubt that splatter is a real issue with the stock Retro, but with an 'afterburner' it'd defiinitely be noticeable. As a relative newcomer to AM, I'll surmise that at higher power levels, the keep-alive is useful in keeping the AGC in listeners' receivers from 'pumping' between words.   

Gentle correction always welcome. <g> I'll be mulling over this whole thread and making changes to the Retro as appropriate. Good stuff! (and more thorough than my own work on this project). There were at least 4 different versions of the board over about 5 years, punctuated by a household move and a slow-motion family crisis.  At some point, the 'better is the enemy of 'good enough' philosophy took over- I wanted it released during my lifetime! :)

73- Dave, K1SWL

[W1RKW]

Hi Dave,
Welcome to the forum!  What took you so long  .

We've had several interesting and fun discussions on the Retro75 over the last year.  There are other threads lurking about too.

Several have contributed much in the way of potential mods and frequency response adjustments.

It's a fun little rig.  You've got a good design there. And it's perfect for experimenting too.

Bob

[kg8lb]

  I really like the little bugger pretty much as-is. The receiver alone is well worth the price of admission. If a better version comes along, I may consider updating. Really looking forward to the rumored 40 meter version . A stand alone receiver based on the Retro , perhaps with DDS tuning could be a real treat.

[WD5JKO]

  Dave,  Welcome to AM Fone. Your little rig has been the topic of much discussion on this board. I wonder how many have been sold? I am sure having fun with mine.

  I finally got back to mine list night taking another look at the need for the 2- diode limiter with keep alive supply. Running 1Khz audio tone at 120% + modulation, the limiter is pretty transparent with the 2nd, and 3rd audio harmonics (as seen on Icom R75 looking at sideband harmonic levels), but I see benefit where the 5th harmonic is attenuated quite a bit. Adding a LPF to the modulated B+ seems to help reduce the 3rd harmonic as well. The LPF was 220 ohms in series with 5uf from the modulated B+ to ground. Still a work in progress.

  The modulation transformer rewiring into a tapped Heising reactor kind of nuked the AGC circuitry as designed, so I have been working on that. Currently have a clip lead storm sensing the modulated B+ through a 20v zener diode, another diode, some more resistors (divider) going into the LED. It seems to kick in at around +100% with a sine wave, and yet still allow 120% peaks with voice. Seems that momentary overload during the attack time is inevitable, and increasing the decay time makes the attack time issue worse on the first syllable spoken, but better thereafter. Still a work in progress.

  I hope to increase the maximum average modulation potential with voice from ~ 60% to 90%, and if successful that doubles the sideband power. On a QRP AM rig, a 3 db increase in sideband power is huge.

  I will furnish more data when I get closer to a final configuration.

Jim
WD5JKO

[WD5JKO]

Ok, I just got off the air with my first QSO (4 stations) with the Retro75! Got decent reports considering the power level using my home made OCF Windom antenna. The receiver seems a little hard of hearing until the stations get on, then I was jumping to turn down the audio! I had a pleasant experience with this thing.

The audio is still a work in progress, and spilling out with clip leads, and a proto board. I need a good day to neaten everything up, and make a little perf board for the transmit audio AGC circuit. I have an adjustable pot to adjust the AGC threshold, and I can limit from about 80% to beyond 130%. The major darn with the circuit as-is is that the pot setting must change with supply voltage to maintain the same mod % AGC threshold.

I attach a hand drawn highlighted schematic, somewhat crude, and Rev. A. Any components with a value are either a change, or an addition. Also attaching a scope plot using speech while saying Yeaaaaaaaaaa where the upward peaks hit 130%.

I switched the dynamic mic phasing, and it was backwards before. The asymmetry is now bringing the positive peaks higher.  

Kind of fun watching the thing super modulate upward without seeing the baseline "white-line". The AGC limiting transistor Q10 acts as more of a clipper than a attenuator, so when I see the peaks clip from AGC action, it is not from the modulator clipping, it is Q10 at a low level clipping the positive peak. This might be bad, but it sounds decent, and clipping like this definitely brings up the average modulation percentage, which in a QRP rig is most desirable.

At 14V DC input, I see about 4W output, and ~ 20W PEP.

Been up all night, time to QRT.  

Jim
WD5JKO

[kg8lb]

Good story Jim,
  Thanks for the updates . Sounds like you have been pretty busy. My first evening of operation netted a roundtable rag chew Utica, MI to Knoxville TN to South Carolina to PA. I got excellent signal and audio reports . Over the course of the QSO I offered to increase the power and was told there was no need , copy was Q5 . My antenna is a 1/2 wave co-ax fed dipole @35'. Since then I have had over 100 QSOs  many of them at 200 miles plus. Quite a few were Retro to Retro. Quite a ball to sit out in my backyard shack running purely on battery power and carrying on a 1 hour rag chew with a ham buddy over 100 miles away !  The receiver in the Retro is indeed surprising. Same here on the low noise yet even the very weak signals seem to leap right out of the speaker.
  Nice thing about kits like this is the low risk factor regarding mods. I have beem content however to run it as-is for now. I have been getting the rated 2.5 watts @ 12.7 volts . The hottest running component is the modulator IC but it is well within a reasonable heat range. The PA is just a bit below that. I had considered a Hammond xfmr upgrade but I will save that for Retro #3. Retro #1 is here and I built #2 for my buddy Mike, N8SDD. Mike is also in awe of the performance of the very simple receiver.
 I am taking notes on your work and will let you know as we progress with #3. I already have the kit but taking time to work on some other projects.

  TNX again, Gary