Print

[VK6HP]

HH, it's true that scattering parameters and associated formalisms are much better developed today but the essentials of stability analysis and synthesis were well in place in 1960.  I'd speculate the real 30L-1 story is probably closer to the one I mentioned earlier, namely that, relative to (say) the 30S-1, there was less high-level engineering involvement at the early phases although I don't doubt that there were heavy duty later efforts to remedy some of the design issues.

One could certainly do a comprehensive measurement of the amplifier but it goes beyond simple, small-signal scattering parameters.  That'd be an OK place to start in order to gain some insight but large-signal (incorporating non-linear behaviour) measurement and analysis, with flexible measurement and reference plane transformations to properly account for external samplers etc., would be more than a wet afternoon's work.

Jim, I'm interested in the source of the numbers for the 30L-1 with directly grounded grids. Could you point me to that?  I've happened across comparisons for other amplifiers but not the 30L-1.  As I mentioned, my amplifier has the original configuration (not directly grounded grids).  Just out of interest, the FL2100B I mentioned also had its original circuitry, not directly strapped grids.

I don't find it unusual that, with a 70W drive specification to achieve full plate input, most 30L-1 users actually get to that with nearer 60W drive, assuming good tubes. I've always found Collins specs to be conservative.  By the way, I can drive the 30L-1 at 70W and get more than my 600W output into the dummy load but there's no absolutely no point, in my view.

73, Peter.

[G3RZP]

Peter, VK6HP,

For what it maybe worth - possibly not much - a story I was told on a professional visit to Rockwell Collins in the mid 1980s was that the 30L1 was originally built at home by one of the Collins engineers, and he took it into work to do some tests. it was seen by management - maybe even Art himself - and got turned into a commercial product.

How true this is, I know not; it is merely what I was told by a Collins development engineer who was also a ham.

The 30L1 has a reputation for marginal stability on the upper HF bands and it is noticeable that, unlike the RCA Application circuit for a four times 811A GG amplifier, it has no neutralising. I find the low value of PA feed choke interesting: there has to be a fair amount of power lost in the choke considering its low value.

73

Peter G3RZP

[VK6HP]

Thanks Peter

Regarding the 30L-1 origins, I have heard and read various comments along those lines. Being far removed from the time and place, about all I can say is that the version you relate, and the consequences, are at least consistent with what I've observed in similar development activities. Coincidentally, I'm currently writing a book about a particular mega-science project and I've come to understand some of the vagaries of personal recollections and oral histories. There is rarely one totally correct source, and sometimes written material from archives confounds even the most sincere individual.  Humans are good at interpolation, justified or not.

As I mentioned, none of this detracts from my 30L-1 experience: it's currently my daily drive while between other amplifiers and, if the new LDMOS* model lasts and performs half as well as the 30L-1, I'll be very happy.

73, Peter.

*heavily over-specified for cleanliness, naturally.

[HAMHOCK75]

Non-linear, high power vector network analyzers have been available for a decade or so,

https://www.keysight.com/upload/cmc_upload/All/HighPwrNVNA-MTT2010-Slideset-100414.pdf

http://literature.cdn.keysight.com/litweb/pdf/5990-5039EN.pdf

They were developed for essentially one customer, the Dept. of Defense.

Where I worked, the code name for a home project using company parts was "G-job". It was encouraged by management as way for people to develop their skills. I do know of one "G-Job" that did become a production product so the Collins story related by G3RZP could have been a Collins "G-job". "G-job" stood for government job if I recall correctly, meaning don't ask.

I was working on a G-job when I went the a company computer club meeting in June, 1975. I met another person working on a G-job there. He gave me a signed and dated ( 6-22-75 ) schematic of his G-job. The name on the schematic is Steve Wozniak. Yes, I still have it.

[HAMHOCK75]

I had been reviewing some MIT, Berkeley papers on amplifiers.

https://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-776-high-speed-communication-circuits-spring-2005/lecture-notes/lec18.pdf

http://rfic.eecs.berkeley.edu/~niknejad/ee142_fa05lects/pdf/lect26.pdf

These papers discussed efficiency. It occurred to me that after the accuracy of the high voltage and current meter had been calibrated, it might be interesting to do some efficiency tests just to see how the AL84 would fare.

Here are the results,

   FT100D      Vp Ip   Vppout   Pin   Pout   Efficiency %

        0         972   0.26   140      253    50      20
      10         948   0.34   180      322    81      25
      13         940   0.37   196      348    96      28
      20         923   0.41   220      378   121      32
      30         897   0.47   256      422   163      38
      40         870   0.52   288      452   207      46
      50         842   0.58   312      488   243      50
      60         827   0.63   332      521   275      53


FT100D is the power setting of the FT100D ( which is not very accurate )
Vp is the measured power supply voltage ( used a Fluke 77 to measure on the power supply side of the RF choke to the tubes, not the front panel meter to get both readings quickly )
Ip is the plate current read from the front panel meter.
Vpp is the peak to peak voltage measured into the dummy load
Pin is the product of Vp * Ip
Pout is the calculated power of the Vpp delivered to the 50 ohm dummy load
Efficiency is ( Pin/Pout ) * 100

A class A amplifier would have efficiency fall directly as a function of output power, where the efficiency of a class B amplifier falls directly as the collector ( plate ) voltage. As an example, if we reduce power from 275 watts to 96 watts the efficiency of a class A amplifier would fall from 53% at 275 watts to  (96/275) * 53% = 19% at 96 watts.

A class B amplifier over the same power range would drop (196/332) * 53% = 31%. The measured efficiency at 96 watts is 28%. The mid point between 19 and 31% is 25% so the AL84 seems a bit closer to class B than A.

It is said that SSB audio has an average power level about 25% of the peak power. If the peak power were set to 300 watts, the average power would be around 75 watts. That does not seem to bode well for four tubes with a total plate dissipation of 120 watts when the 25% efficiency at 81 watts output power requires 322 watts input.

[VK6HP]

Re G-jobs...

We call them "FOs" : "foreign orders".  I once fired an engineer who was so inefficient he was behind in his own FOs

I've seen the Keysight (and some competitor) large signal VNA systems demonstrated at industry briefings but, despite having access to a very good range of exotic gear at the low-noise, small-signal end of the application spectrum, we sadly have little high-power exotica.  If you can get your hands on a high power system, by all means go for it.

[HAMHOCK75]

Have you ever seen these books by chance?

https://www.amazon.com/Design-Amplifiers-Oscillators-S-parameter-Method/dp/0471092266

https://www.amazon.com/Low-Noise-Electronic-Design-C-Motchenbacher/dp/0471619507

[VK6HP]

Motchenbacher and Fitchen, and a later volume, I've used as a reference for graduate students, and for my own general reference. Can't say I've seen the other one.

There are some confronting lessons from my own field of radio astronomy where, for a long time, cryogenically cooled devices and LNAs were the norm.  In general, we aimed for the very best noise temperature attainable and if an occasional LNA (probably using bleeding edge device technology) had to be babied into stability by one-off bias or other adjustments, that's what we did.  We still have systems like that but these days, with big antenna arrays or wide field-of-view focal plane arrays within antennas, or combinations of both, we have a much more production-like environment, usually based on high-performance ambient temperature systems. Stability is much more of an issue.

It's extra complicated because, as well as very low noise in the feed/antenna + LNA combination, astronomers often want very-wide bandwidth, and very low cost.  Sometimes those "verys" have rolled together to produce deployed systems with stability issues.  These days, noise wave and associated circuit modelling get us closer to the mark.  

But maybe I'm more tolerant than some of the 30L-1 stability tribulations

73, Peter

[HAMHOCK75]

Does one of the Motchenbachers have an appendix with a chart showing dBc vs L(f) for ECL and TTL dividers or a page with an amplifier using possibly a 2N6429?

[VK3BL]

It is said that SSB audio has an average power level about 25% of the peak power. If the peak power were set to 300 watts, the average power would be around 75 watts. That does not seem to bode well for four tubes with a total plate dissipation of 120 watts when the 25% efficiency at 81 watts output power requires 322 watts input.

Wow HH75, I'm amazed. 

Someone actually gets it that 'SSB' is not low duty cycle by any means.

Low duty cycle to an 811A amplifier is CW DXing.  Thats it, a few '5NN TUs' here and there.

[HAMHOCK75]

I was surprised by the data. For some reason it never occurred to me that the average SSB power was occurring at such a low efficiency. I also just realized that I uploaded the wrong Berkeley paper. This is the one that was intended.

http://rfic.eecs.berkeley.edu/~niknejad/ee142_fa05lects/pdf/lect27.pdf

[HAMHOCK75]

Looks like the transformer was not just loose. Some of the mounting ears had cracked and broken off. The screws are not easy to get out. The holes are filled with what appears to be some black dip material. Interesting that the washers under the heads of the screws are fiber washers not metal maybe to allow for some expansion.

[HAMHOCK75]

There are some confronting lessons from my own field of radio astronomy where, for a long time, cryogenically cooled devices and LNAs were the norm.  In general, we aimed for the very best noise temperature attainable and if an occasional LNA (probably using bleeding edge device technology) had to be babied into stability by one-off bias or other adjustments, that's what we did.  We still have systems like that but these days, with big antenna arrays or wide field-of-view focal plane arrays within antennas, or combinations of both, we have a much more production-like environment, usually based on high-performance ambient temperature systems. Stability is much more of an issue.

This sounds very familiar. In everyone's career there are Y's in the road. Sometimes you wonder where the other road would have led. I had a chance to work on a National Science Foundation grant or SETI with this person

https://www.seti.org/seti-institute/about-seti/bernard-m-oliver

SETI being https://www.seti.org/centerforseti

I vaguely recall it was a huge array of antennas but someone needed to calculate the sensitivity of the array to all the different length cables connecting them all and the consequences.

I ended up working on this National Science Foundation grant. My advisor was Ward Helms. Ward suggested that if I didn't find a job, I could continue for the PhD. My next stop would have been New Zealand, then Navy transport to McMurdo station in Antarctica, then a winter at the Longwire substation in the photo on the Ross ice shelf to collect data using a 25KW transmitter and a 21 mile long antenna. I found a job, lol!

https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=2&cad=rja&uact=8&ved=0ahUKEwjm5PvExpzZAhVO62MKHeeMCsYQFggpMAE&url=https%3A%2F%2Fs3.amazonaws.com%2FAntarctica%2FAJUS%2FAJUSvVIn1%2FAJUSvVIn1p21.pdf&usg=AOvVaw20HUwOSE6dmyOuTVJ_-iz9

[HAMHOCK75]

A trip to Home Depot yielded 1/2″ bar stock, threaded 8-32 rod, 8-32 nuts and fiber washers. They didn’t have 8-32 screws over 3″ long but they did have 8-32, 3 feet long threaded rod. They also did not have 1/2″ x 1/16″ steel bar stock but did have 1/2″ x 1/8″ steel bar stock. The rod and steel stock were easily cut with a reciprocating saw. A propane torch was used to heat the steel bar stock before bending in a vise. The holes in the transformer were cleared with a 11/64″ drill bit. Left to do are to drill holes in the mounting brackets, polish with a grinder and paint black.

[KM1H]

The 30L1 has a reputation for marginal stability on the upper HF bands and it is noticeable that, unlike the RCA Application circuit for a four times 811A GG amplifier, it has no neutralising. I find the low value of PA feed choke interesting: there has to be a fair amount of power lost in the choke considering its low value.

And a well deserved reputation at that and even when almost new and tuned by non technical USAF MARS station operators. Ive related my side trips on here to those sites while in the USN cruising the Med in the early 60's.

I wonder if that plate choke as well as ones in Heath SB family amps had an effect on stability on any band. I do know that an Ameritron Choke in a SB-220 settled it down considerably.

Carl