The AM Forum

For tube geeks,

I found the attached vacuum tube broadband amplifier chart in my father's technical archives yesterday.  It plots tube capacitance versus transconductance for the old tubes and computes the gain-bandwidth product by the formula in the lower right corner of the graph.  There is no note telling what the source is and I have never seen this graph anywhere before as I recall.  

The graph is 14" long, so I scanned it in at both ends, print and tape. I think there exists a utility to fuse these two scans together if that is a help to some.

If you do print out the two pages, make sure you DO NOT select ''Fit to printable area" so that both pages are the same scale.

You probably should have scanned them in the same direction. Scanning in opposite directions generally will throw the registration off when trying to bond them together.
When you print them, you really can't get most of the lines, numbers, and prints to all get lined up even close.

I do this all the time for large schematics that are horizontally longer then 19 inches.

Hi Pete,

I have seen that pages don't match up on some boat anchor receiver schematics I have printed out in the past. When I printed out this chart the 2 pages matched perfectly after not selecting 'fit to page'.  I am using a Brothers laser printer and a Canon flatbed scanner.  Looks like everything here is linear. 

I guess the old scans may have been non-linear and perhaps the receiver schematics I am referring to were printed on my old (now dead) ink jet printer (another source of non-linearity?).  It seems that all of this stuff is stepper motor driven and so should be linear.  Some scanners may have had some optics involved, certainly a source of non-linearity.

I needed a slight change of pace today. Seems like I'm doing manuals 7 days a week  ;D
See your charts below.
One is roughly 11X17 and one is 8.5X11 for printers that don't have the ability to print on bigger sheets.
I make old stuff look good  :D

Thanks Pete. Yes, looks real nice!  It was on a piece of paper that looks like a blue-print almost. 

I wonder where the graph came from.  I was thinking from a book in the Rad Lab series at first, but there are post war tubes on the chart.

Also looks like the original paper is acidic. Probably in another 20 years or so, you won't even be able to read the print since the background will just get darker.
Fortunately, the machine I use here, can really dance well with this type of acidic paper problem and really clean up a lot of this stuff. Some original military manuals I have here, the sheets have all turned brown, but run through this machine, no background  sheet color comes out.

"If the GBWP of an operational amplifier is 1 MHz, it means that the gain of the device falls to unity at 1 MHz. Hence, when the device is wired for unity gain, it will work up to 1 MHz (GBWP = gain × bandwidth, therefore if BW = 1 MHz, then gain = 1) without excessively distorting the signal. The same device when wired for a gain of 10 will work only up to 100 kHz, in accordance with the GBW product formula. Further, if the minimum frequency of operation is 1 Hz, then the maximum gain that can be extracted from the device is 1×106."

Also:

"Power Gain and Bandwidth of RF Amplifiers

Power gain and bandwidth of RF amplifiers depend on both device and circuit properties. As noted above, the power gain of a tetrode falls in the range between 30 dB at LF and MF and 10 dB at UHF. Power gain at UHF is limited by electron-transit-time effects and by the inevitable effect of cathode-lead inductance.

For a simplified ideal tube amplifier, the gain-bandwidth product tends to be constant and equal to Gm/C, where Gm is the mutual conductance (transcon-ductance) and Cis the output capacitance. This product is independent of frequency. In practice, the product is always less because of the presence of the external circuits that store energy and because the bandwidth of the input circuit may be smaller than that of the output circuit. Furthermore, the performance of the amplifier may be largely determined by the amount of feedback present. It is nevertheless true that the gain-bandwidth product will always increase with Gm/C."


Helps to explain something about the chart...

Not sure that tubes, by themselves actually exhibit a GBP??
(unless they are characterized similarly to transistors with fT??)


                      _-_-

Well I thought it was an interesting graph to look at Bear, the context, who knows? 

About 80 tubes are plotted, many of the common tubes used in receivers are there, across many families.  I see 34, 77, the 6C6 and 6D6 – 6-pin tubes from around 1935, octals, loctals, miniatures, acorn.  The 5847, one of the 3 on the right-end of the graph is better known as the WE 404A, no surprise it’s where it is.  Bear, is that what tube was in your multi-coupler distributed amp?

Yeah, I don’t think you can put a GBW on a tube like with op amps.  A current pumped into a capacitor gives a voltage slew, but I think some engineer wanted to have a figure of merit for the tube comparisons and that’s why the simple formula is shown on the graph.  No designer looking at post-war tubes is going to choose a 1930’s tube. Perhaps the graph was originally done in the war and then updated post-war?

As far as op amp circuits, normally you want the op amp to have 10 – 20 dB of open-loop gain compared to the closed-loop-gain at the application frequency, but depends on the application gain-error criteria. I would not want a 1 MHz GBW op amp voltage follower to have to handle a 1 MHz signal.  I haven’t looked at the open-loop gain curve on today’s op amps, but although it follows 1/f, I think they usually plateaued around 10 Hertz. 

But we’re getting off of the topic of a unique, slightly bizarre, tube performance presentation.  Unfortunately my father, a tech writer, has been a SK for 10 years so we cannot ask him where the graph came from!  (Hey, the 6AK5 sure looks attractive!)

This is a nice chart. A similar one(s) exists for solid state like
FTAU (ft) which is the unity short circuit current gain crossing point. The
equation you obtain for that process is a similar formula absent the output
capacitance, ft=gm/(2pi*Cin), Cin is Cgs for a FET.  I should add, ft is not a bandpass response measurement. Its response extends to DC and it is evaluated for a single device. You can measure the low frequency current gain of a transistor, say with a 50 ohm source and load and extrapolate where that gain goes to one. Not like the circuit response provided by the equation in the tube chart which is based on a bandpass system.

The result for the tube calculation is a bit different in both the input and output
circuits are assumed to be tuned. Terman does an excellent job in a paragraph
to explain. In wideband RC coupled amplifiers with no compensation tweaks, the
largest possible bandwidth would be the one in which the input and output would be
resonated with an LC circuit. It is desired to use the largest L possible so as to not
limit the high frequency roll off . That is, little shunt C. Hence, the lowest possible total C is due to the tube itself, Cgk and Cgp. Gm comes about since the tube gain transfer is output plate current shift per input grid control voltage shift... dIp/dEg, or gm. Any input C present in the tube, like the transistor, provides a go around path for the control voltage and the gain DROPS.

This GBW product is really a figure of merit for tubes and solid state. If a 6AK5 for example is used with a GBW merit of 120 MHz, then it should be possible to squeeze 12 dB of voltage gain over a BW of 10 MHz in an RC coupled amplifier. True, if ZERO circuit capacitance is present.

Alan