Bootstrapping

[kalee20]

Originally Posted by dinkydi

You are quite right that the input impedance would be R1 alone. In a real design you would remove R3 because it wastes loop gain and R1 & R2 can supply the grid current (or R2 alone if R1 is ac coupled).

Although quite often, R3 is needed simply as a necessary DC grid leak!

[dinkydi]

Hi kalee20,

With shunt feedback, as described, R3 does not operate as a grid leak in the accepted sense because of the influence of R2 (and R1 if not ac coupled). Grid-leak biasing and shunt feedback are incompatible unless you run the stage open-loop at low frequencies. This may be useful for filters/equalisers but is OT for this thread.

Peter

[kalee20]

Originally Posted by dinkydi

... unless you run the stage open-loop at low frequencies...

True.

Bootstrapping is one of those things which can be a bit hard to pin down, but you know it when you see it.

There are two distinct reasons to bootstrap - one being to raise an impedance (common with transistor emitter followers, but I have also seen a pentode voltage amplifier directly coupled to a triode concertina phase splitter. The triode is already effectively infinite impedance, but the clever trick is that the cathode is capacitively coupled to a tap on the load resistor of the pentode, thus the pentode works into an effective load many times the load resistor value).

The other is to get a wider voltage swing - this used to be common in transistor audio amplifiers where the output was coupled via a capacitor to a previous stage load, thus allowing this to swing outside the supply rail and driving the output stage with maximum possible signal.

[daviddeakin]

Originally Posted by G8HQP Dave

There is another valve bootstrap circuit which is more like the bootstrap often seen with transistors. This is the grounded cathode followed by a cathode follower, with the first stage anode load boostrapped from the CF output. It is closely related to the SRPP (and therefore a cousin of the mu follower). Does anyone know when/where this circuit first appeared?

The SRPP is just one of a family of circuits that were all once referred to as 'bootstrap amplifiers'. The first SRPP was patented in 1940, as far as I know.
Clough, N. H. Improvement in or relating to Modulator Arrangements. British patent 526418

However, the mu-follower, which is another member of the bootstrap family, appeared in the 1930s.

A triode loaded by another triode as an active current source (which is the progenitor of all those circuits) goes back even to 1922!
http://www.google.com/patents/US1403...403566&f=false

[kalee20]

Originally Posted by daviddeakin

A triode loaded by another triode as an active current source (which is the progenitor of all those circuits) goes back even to 1922!
http://www.google.com/patents/US1403...403566&f=false

I've just had a look at this circuit, but it's not a cascode - the 'load' triode has its own bias battery between cathode and grid. So the input triode isn't feeding into a grounded-grid triode - the grid is not at a fixed potential with respect to circuit 0V.

The second triode is effectively just operating as a 2-terminal device - I'm too lazy at present to read all the patentese, but it is difficult to see what it's doing that couldn't be done with a resistor.

One could say the second, 'load' triode has its grid bootstrapped by the battery, but I'd argue against this, as the battery is forced up and down by the stage input, not by the output.

[Refugee]

The two terminal wired triode will be operating in a way that produces a constant currant.

[kalee20]

I agree Refugee - effectively it's an active load, allowing higher gain than a resistor will give.

Thus anode voltage swing of the first stage will be maximised - the opposite of what a cascode would do.

[daviddeakin]

Originally Posted by kalee20

One could say the second, 'load' triode has its grid bootstrapped by the battery, but I'd argue against this, as the battery is forced up and down by the stage input, not by the output.

The cathode resistor of the upper triode is the thing that is bootstrapped. Its value is multiplied by the triode's mu.

[Refugee]

Kalee20 has spotted that i have just been tinkering with a bench PSU that has a silicon copy of the device that has been found.
The beast had to come apart after the constant currant bit had been tested and it can be seen in the success story section, just look for for Wier 423D with the search button at the top.

[Radio Wrangler]

Constant current sources can be very useful design tools.

The lower valve sees a very high impedance loading it, but its own anode characteristic curves have a voltage versus current slope, which sets an unavoidable self-load, which limits the available gain from the stage. It is effectively in parallel with whatever external loading is connected to the anode. The Gm transconductance times the anode slope impedance sets the mu, the max available voltage gain of the valve if worked into a true infinite load.

In reality, the upper valve, the constant current source is imperfect, it too has an anode curve slope, and so a limit on the impedance it can present. If the two valves are the same, then the gain is going to be limited to about half the databook mu value.

Tetrodes (and pentodes and onwards) by screening the grid region from the anode field make the anode characteristic slope less, or in other words the self load impedance bigger, and this is what allows them to have greater gain if the external load impedance allows it.

Look at transistor curves and the Ebers and Moll model for example and you'll see the same things happening. There's not much new under the sun...

David

[G8HQP Dave]

There are two advantages of the 'active load' circuit:
1. If both upper and lower cathode resistors are bypassed (or both unbypassed) you get some 2nd-order cancellation as the upper valve and lower valve curve in opposite ways. Gain is mu/2 in this configuration.
2. If the lower cathode is bypassed and the upper not, then the lower valve sees a much greater anode load so can give greater linearity. Gain approaches mu. The active load done this way is better than a simple resistor.

[Synchrodyne]

I recently happened across another valve-era reference to bootstrapping. I was browsing through “Radio and Television Engineers’ Reference Book”, Third Edition, 1960 with no particular intent when I came across, in Section 47, Progress and Developments", an item on GEC colour TV receiver developments (NTSC, 405), which was about the experimental TT4 (sic) model.

Included in the description was: “The luminance signal is amplified by two video stages in a “bootstrap” circuit...”

As it happens, Carnt & Townsend, Volume 1 included the luminance circuit for the GEC TTIV, copy attached, although there is no mention of bootstrapping in the accompanying text.

Looking back from the N709 luminance output stage, I think that the LZ319 (pentode section) pre-stage would appear to be a bootstrapped cathode follower, although it also seems to be acting as a more normal video amplifier with some gain to drive the sync separator and agc circuitry from its anode.

Cheers,

[kalee20]

I'd argue that the input is applied between cathode and grid (easy, because it's a floating, transformer-coupled input), and the output is taken between cathode and anode (with the power supply in between).

So it's a common cathode amplifier, albeit with the load in the cathode and the anodetaken (nearly) directly to HT - that's ignoring the auxiliary output. Effectively the anode load and power supply have been swapped around.

What is interesting, is that the valve is operating as a triode! Or is it? If the RFC's in the screen-grid and also the anode, are small, then the anode is effectively connected to the screen. On the other hand, if the RFC in the screen has significant impedance, then screen voltage will vary - but in that case I'd expect a decoupling capacitor to exist, between screen-grid and cathode (not chassis).

[Synchrodyne]

Looked in an overall sense, and not just from the perspective of the N709 output stage, this circuit might be thought of as being of the distributed load type with a bootstrapped input.

Cathode loading implies screen grid loading. Assuming that the choke in the anode circuit is for peaking purposes, and mostly for the benefit of the main (cathode) output, then the screen would need comparable peaking to ensure that all of the cathode current saw the same net effect.

By the conventional wisdom, distributed loading means the valve is operating as a cross between a triode and a pentode.

Actually, one is reminded of the Quad output circuit, although in that case the cathode and anode were driving the same ultimate load via the output transformer, whereas in the GEC TTIV case, the cathode and anode loads were separate. In the early days at least, Quad described its output circuit as operating partly in the cathode follower mode.

Clearly, the GEC circuit would have less then unity gain at the cathode output, but I should expect quite a bit of gain at the anode output, enough to drive a valve sync separator.

Cheers,

[G8HQP Dave]

Originally Posted by Synchrodyne

Clearly, the GEC circuit would have less then unity gain at the cathode output,

Depends how you define gain. I would say it has quite high gain to the cathode, given that the input is from cathode to grid but the main output is from cathode to ground.

[kalee20]

Originally Posted by G8HQP Dave

I would say it has quite high gain to the cathode, given that the input is from cathode to grid but the main output is from cathode to ground.

Yes I would, too. Certainly it has the potential to have much more than unity voltage gain.

[Synchrodyne]

Thanks for the correction. Yes, I had forgotten that the input is not grounded on one side, but effectively floats above the cathode output. Thus the gain of the latter is not constrained by the usual cathode follower rules.

Cheers,

[daviddeakin]

With some irony it appears there is no bootstrapping in that circuit at all! It is a common-cathode amplifier, followed by a common-anode amplifier, followed by another common-cathode amplifier.

The second stage could easily be confused with a cathode follower, and the brightness pot sort of makes the third stage look a bit like a bootstrapped amplifier because of the way it's drawn- but it isn't!

[Synchrodyne]

OK, the luminance output amplifier (N709) is a conventional common cathode stage, with variable negative grid bias to set the black level (brightness control).

I had assumed that the bootstrapping term was applied by GEC to the first luminance amplifier, ½LZ319.

In the attachment, I have drawn out what I think are simplified versions of both a regular bootstrapped cathode follower and the GEC circuit.

In the regular circuit, the input signal is effectively applied across R1 and R2 in series. But owing to cathode follower action, point B, the top of R2, is very close to point A, the top of R1, in terms of signal voltage. Thus the bridge effect means that the current flowing through R1 is miniscule, in turn meaning that as far as the input signal is concerned, R1 appears to have a much higher impedance than its rated value. Were the bottom of R1 connected to earth, as in a normal cathode follower, the input would see R1 as having its rated value. Thus the bootstrapping, which is the connection of the bottom of R1 to the cathode follower output, gives rise to the apparent impedance multiplication.

In the GEC TTIV circuit, the input voltage (from the final IFT secondary and via the demodulator diode) is applied across R1, not between the top of R1 and earth. This makes B the common cathode point of what is apparently a common cathode amplifier. Output signal voltage appears across R2. As the gain could be higher than unity, signal voltage at point B would not necessarily be close that at point A. It would be the case, though, if the gain in terms of signal voltage across R2 were unity. So in general the GEC circuit does not have a bootstrapped input, but it does appear to be bootstrapped in the special case that the cathode output is unity.

Perhaps it did have unity gain at the cathode output, and that is why GEC referred to it as a boostrapped circuit. Or perhaps GEC confused itself simply because at first glance, it looked like a bootstrapped cathode follower.

Conceivably the former applied. It seems possible that the N709 would produce enough luminance output voltage swing from a normal (say around 3 V p-p) vision diode demodulator output. Thus the ½LZ319 served as a unity gain buffer in the main luminance channel, presenting a high impedance looking back to the demodulator and a low impedance looking forward into the delay line. At the same time it provided enough gain from its anode circuit to drive both the sync separator and agc circuits.

Cheers,

[Synchrodyne]

Recently I came across a Wireless World 1961 January article, “The Bootstrap Follower”. The article is available on the web at: http://www.douglas-self.com/ampins/w...hive.htm#bootf.

Essentially this article reviews the circuit hitherto described here as the bootstrapped cathode follower and notes how and where it differs from the conventional, or true cathode follower. The author (G.W. Short) proposed the name “bootstrap follower” for this circuit, after deducing that in practice it was like a hybrid between a cathode-follower and a bootstrap amplifier. The latter had been described thus: “...a triode amplifier with a floating input voltage and cathode output is a bootstrap amplifier,...”

That takes us back to the GEC TTIV luminance amplifier, sketched out in the previous post. This, with its floating input, would thus appear to satisfy the definition of “bootstrap amplifier” given in the WW article, and from the context of that article, one may reasonably infer that that was a definition in common use at the time. So GEC may well have been following common practice of the era in describing its luminance amplifier as using a “bootstrap circuit”.

I am not sure if this helps to clarify or just adds to the apparent confusion, but in any event the additional historical information does seem to be germane.

Cheers,