How reliable is a TTL gate as linear amplifier?

[RoGeorge]

Sometimes, CMOS inverter gates are used with as linear amplifiers, but I need to do that with a TTL instead of CMOS.  I've seen many times CMOS gates used in an analog manner, with a negative feedback, but I didn't see that too often for TTL gates.

There are some TTL examples, too (e.g. https://electronics.stackexchange.com/questions/284741/ttl-nand-analog-amplifier-does-not-work-in-multisim-why), yet they are not that many, so my guess is this is because the outcome is not as reliable as with CMOS inverters.


Maybe the PSF is not stable enough with temperature (TTL get warm while CMOS don't)?
Or maybe it's because of the multiple slopes in the TTL gate transfer function? 


Source:  https://wiki.analog.com/university/courses/alm1k/alm-lab-27

Asking because I would like to use an open collector NOT gate SN7406 as an audio amplifier for an AM modulator (so no Hi-Fi, just MW radio quality).

I would prefer the TTL gate only because I happen to have a 7406 in that schematic already.  It's a one-off project for the fun of it, not designed for manufacturing.

My question is, did anybody tried using a TTL gate as audio amplifier?  Is it reliable enough, or should I just use a normal amplifier?

[jpanhalt]

I came across another version just last Friday.  It works, but I have no idea about its reliability.  Why would it fail?

[RoGeorge]

The CD4049 in the attached gauge.PDF are CMOS gates.  Mine are TTL (the internal schematic of a TTL gate is different from a CMOS gate, TTL is using bipolar transistors, CMOS is using complementary MOSFET transistors).

In theory, a TTL should work as an amplifier, won't fail with smoke, but I suspect it might behave unreliably, for example because of thermal drift, or because of multiple slopes non-linearities specific to TTL gates, or maybe because of something else.

In practice, schematics examples with TTL gates as linear amplifiers are rather rare, relative to similar CMOS schematics, so I guess that's because they do not perform well.

That's why asking about the success rate with TTL gates as linear amplifiers.

[srb1954]

A major problem with trying to use a TTL in a linear circuit is that the input sources significant current that changes dramatically as the input voltage changes. On top of this any variation of the threshold voltage and change of transistor beta with temperature makes it difficult to reliably define, with simple resistive biasing, any sort of stable operating point for linear operation.
 
The old Texas Instruments book "Designing with TTL integrated circuits" is worthwhile reading as it gives a lot of useful information on the internal structure of TTL gates and the considerable variation of their parameters over temperature and supply voltage.

N.B. I doubt that the bias voltages shown on your attached diagram are correct. With the input at +4.5V DC that input stage would be totally cut off and the output voltage would be stuck hard low. For this circuit to operate as an amplifier the input bias voltage would have to be somewhere around 1.4V for the input stage to be operating in its linear region.

[RoGeorge]

Good catch about the 4.5V in the scanned schematic.  Most probably a 1.5V printed in the book as 4.5V because of a typo error.

[voltsandjolts]

Perhaps relevant and useful:

TI App Note SZZA043: Use of the CMOS Unbuffered Inverter in Oscillator Circuits

It discusses choice of bias point and dependence of open loop gain on Vcc.

[magic]

If it's anything like a TTL NAND gate with one input and the high-side output driver omitted, then it's basically a single grounded emitter NPN, with an emitter follower in front of its base (or is it omitted too?) and a diode half-bridge voltage follower in front of that.

You need to sink the pullup current and then you are driving the base of the output NPN at unity gain with some offset and hopefully not too much phase delay.
Ergo, it may work. Perhaps.

Look up NAND schematic if unsure what I'm talking about.

edit
Quick test: short output-input and see if it stabilizes near one or two diode drops above ground, or oscillates.
If it stabilizes, it's unity gain stable and can do any inverting configuration like unbuffered CMOS. Input voltage will exhibit usual Vbe thermal drift.
If it oscillates, maybe it could still be tamed by raising high frequency noise gain. I think it shouldn't be as hopelessly unstable as buffered CMOS or comparators.

[RoGeorge]

At first, the idea was to use a class C amplifier as both an AM modulator and an RF amplifier, like in this schematic:


Source:  https://ia902501.us.archive.org/12/items/circuit-analysis-theory-and-practice-pdfdrive/Build%20Your%20Own%20Low-Power%20Transmitters_%20Projects%20for%20the%20Electronics%20Experimenter%20%28%20PDFDrive%20%29.pdf

except that I wanted to use only TTL chips, no transistors, so instead of Q3 I've put the 7406 gates, which are open collector, something like this:



Then I've thought I could use less than all the 6 gates in parallel, and instead, use a few of them as the audio amplifier for the final modulating/RF power amplifier transistor(s).

Tried that on a breadboard last weekend and didn't like it.  Too sensitive to the resistor's values, to the supply voltage, and too low input impedance for a typical 47k line in, could work only if connected to a low impedance headset output.  I guess I'll just use a power amplifier IC instead, to drive the final transistor(s).

Thank you all for your advice.