15092	2018-09-25 04:32:06	Andrea Baldoni	IF stages - tube vs. solid state
Message	Date	From	Subject
Hello! While doing internet research for the objective of building a theremin musical instrument, whose components are mostly the same of a superhet as it includes oscillators, amplifiers, a mixer and an envelope detector, I read that it's in general more difficult to obtain the same selectivity in a tuned amplifier (about 455kHz) with solid-state circuits than with tubes. During the development of my (JFET) theremin, actually I had to resort to a ceramic filter (used as slope detector in a frequency-to-voltage converter) because with a simple single LC filter (as used in the original tube based, 1929, RCA instrument) I wasn't able to obtain, by far, a slope steep enough. While probably the reason of low selectivity in my LC was just my fault, again, I read in the book "Build your own transistor radios", that to match the performance in selectivity (and sensitivity) of tube radios, transistor circuits needed more IF stages. I started to question myself why it could be so, and I am asking more experienced builders in the group, opinions, or, if the whole statement that selective IF stages were more easily obtained in tube radios has not any scientific ground, or it had some reason only when BJTs were still in their early days. So far, I understood that JFET Goss and BJT Ro (Early effect), cause a lower-than-infinite output impedance of common-source or common-emitter amplifiers as generally used in tuned IF stages. This reduces the Q of the load. I tried to compare this with a pentode (12BY7A) and, at low plate current (4mA), the plate characteristics graph show 1mA change from 100V to 500V, it's 400kR, actually really better than the g_oss of 50uS (=20kR) of the MMBF5484 JFET I used (but g_oss decreases with Id). In a BJT circuit, Ro also depend by Ic, so let's say both could be about 100kR. I also thought that, miniaturization of circuits caused the Q of tuned transformer to be much lower, and so really the comparison is not between tube active devices vs. solid-state ones, but between whole tube "radios". Just as an idea, a 1945 Minerva radio has a 50x50x95mm IF "cans", while the common Toko 10K transformers are 10x10x12mm, 200 times less volume... Of course with BJT and JFET, it's easy to use a cascode to solve the problem completely, or a tap in the IF transformer so to better adapt to the output impedance of solid-state devices while at the same time avoid Q reduction, but I think will offset the comparison. Andrea Baldoni
15093	2018-09-25 13:15:27	Harold Smith	Re: IF stages - tube vs. solid state
I haven't looked into this, per se. But I have done commercial design work, and the smaller inductors, especially, are significantly lower Q than the ones I pulled out of older radios when I was a kid. I've seen inductors with a maximum Q of 40 described by their makers as 'High Q", which usually would get me laughing out loud. I don't know when the book you mentioned was written, but my memory of the early days of solid state is that it took folks a while to realize just how much lower the impedances in SS equipment were, compared to the tube stuff they were transitioning from. de KE6TI, Harold
15096	2018-09-26 14:02:15	kb1gmx	Re: IF stages - tube vs. solid state
>>>While probably the reason of low selectivity in my LC was just my fault, again, I read in the book "Build your own transistor radios", that to match the performance in selectivity (and sensitivity) of tube radios, transistor circuits needed more IF stages.<<< The low Q parts were and transistor radios are not a case of condition A results in B. Many portables needed more IF not for selectivity but due to the transistors of he day. With two IF stages they typically had 3 tuned circuits where the average 5 tube wonder (AA5) had 4! In both cases selectivity for AM was not much a problem. One of the primary issues of tube radios was insuring less than attainable bandwidth so the AM sideband cutting was not an issue. IT was a bit better with transistor IF but even then they are often loaded with resistors or more heavily loaded taps to insure adequate bandwidth. I know this as i'e used the common trasnistor IF cans to make usable SSB filters (ok, lossy but steep skirts). However for a theremin (I've done tube, Transistor and Jfet) choice of oscillator frequencys and tuned circuit Q are important. Most of the 1929s designs were sub 400khz for both selectivity in not very high Q circuits. Move up to twice that and the selectivity (Q) has to go up a far greater amount and it also makes tuning stability a greater issue. Tubes compared to the modern parts load LC systems somewhat less so comparing a JFET loading a LC and A 12SK7 loading a LC is not a 1:1 comparison even for the same LC. That is whats called loaded Q. So in every system you have Q an Ql. So LC Q interacts with two things circuit loss and bandwidth. So simple comparisons are marginally useful. Allison
15099	2018-09-26 14:34:12	blumu	Re: IF stages - tube vs. solid state
Informative overview, Allison! Many moons ago, I remember experimenting with a superhet with a view to narrowing the IF response for SW listening. I introduced feedback in the second IF stage; this worked, but was, not surprisingly, not stable. I then let the modified IF stage oscillate for SSB resolution; that too worked - sort of. Injecting a BFO signal was better, however all these impromptu mods were somewhat unstable, but indicated that the approach merited further investigation, at the time beyond my means. Michael 2E0IHW On 26.09.18 22:02, kb1gmx@arrl.net [emrfd] wrote: ... > One of the primary issues of tube radios was insuring less than attainable > bandwidth so the AM sideband cutting was not an issue. IT was a bit better > with transistor IF but even then they are often loaded with resistors or > more heavily loaded taps to insure adequate bandwidth. I know this as > i'e used th e common trasnistor IF cans to make usable SSB filters > (ok, lossy but steep skirts). ... > So LC Q interacts with two things circuit loss and bandwidth. > So simple comparisons are marginally useful. > Allison
15100	2018-09-26 16:33:16	Mike Dinolfo	Re: IF stages - tube vs. solid state
Adding feedback to a 2nd IF stage to improve selectivity was sometimes used with the surplus "ARC-5" (vacuum tube) receivers in the 1950's and 1960's. One such approach was described in QST magazine (March 1960) by Lewis McCoy W1ICP "A Poor Man's Q Multiplier", available to ARRL members at arrl.org in the QST archives. McCoy indicated this improved selectivity considerably in the ARC-5 BC-454 (3-6 mhz tuning range; 1415 khz IF) and ARC-5 BC-455 (6-9.1 mhz tuning range; 2830 khz IF). Mike N4MWP
15101	2018-09-26 16:40:16	Gary	Re: IF stages - tube vs. solid state
I remember making that mod to my dad’s BC-454. Remember because that’s where I learned what a gimmick capacitor was! 73, Gary WB6OGD Sent from my iPhone
15102	2018-09-27 00:53:28	Andrea Baldoni	Re: IF stages - tube vs. solid state

15103	2018-09-27 02:48:04	Andrea Baldoni	Re: IF stages - tube vs. solid state

15104	2018-09-27 07:00:16	Andrea Baldoni	Re: IF stages - tube vs. solid state
Hello. About difference in output impedance I just found that suggestion, from a HAM who converted some tube radios to solid state: "Yes you can connect a FET drain directly to an IF transformer made for tubes but the coil will not peak properly. It will be very broad. To peak properly I connect the drain through a resistor (usually 2200 ohms) and a small 2.5mh choke in series. Then I connect a small cap (sometimes as low as 15pf) between the drain and the IF coil." Andrea