Best explanation of how a Phase Inverter works

[robtone]

I have been struggling with how a phase inverter works. In my opinion, this article did the best job of helping me to understand the concept.

http://www.300guitars.com/articles/arti ... -inverter/

I hope it helps someone.

[ChrisM]

Great article, thanks.

[Super_Reverb]

This is a great article. I had never looked at PI design in this light before. If you're familiar with analog circuits, you'll recognize the phase inverter, that we call a long tailed pair, is used universally in op amp front ends.

If you look inside an op amp, the solid state implementation is identical in function, but typically implemented with current sources where we use the large cathode resistor and plate loads.

Known as a differential amplifier, the circuit, with matched devices (BJTs, FETs , or tubes) responds to differences in its input voltage (vg1-vg2).

We typically use it to generate differential output phases because there are no complementary output devices in tube land, i.e. ,PNP for an NPN or P-chan MOSFET for N-chan MOSFET. BTW, if you look at an N-Chan JFET or depletion load N-chan MOSFET, you'll recognize it as a field effect device, which is turned on (current > 0) with no bias and has to be biased slightly negative gate to source to be usable in analog circuits. Does this sound familiar?

There are instances where, instead of connecting the (-) input to AC ground in the L/T PI , we tie in a fraction of the ouput voltage. This is the magic of negative feedback which reduces (open loop) gain, extends frequency response and reduces distortion somewhat.

Merry Christmas,

rob

<edit> p.s. there is a great summary of Pagey's amps during the reign of LZ on this same website.

[mlp-mx6]

He only has one statement in there that I take issue with:

"THERE IS NO SUCH THING AS A PERFECTLY BALANCED PHASE INVERTER!!!"

I believe there is. If you use a center-tapped transformer it is balanced.

[Markusv]

I believe there is. If you use a center-tapped transformer it is balanced.

Except that the windings on the one side of the centre tap may be a bit larger /longer due to coils wound on top of each other?
Am I splitting hairs?

[David Root]

That is the most easily understood explanation I have ever seen.

As to whether or not there is a "perfectly balanced LTP", I think the LTP is a bit like Alice in Wonderland's Cheshire Cat, which mysteriously appears and disappears in different places.

Yes, you can balance the DC voltage pretty easily with a trimpot ala Dumble. But the odds of having a balanced AC voltage reduce. Balance the AC and the DC goes off, and vice versa.

If by some magic you have a double triode with perfectly matched triodes, IE same transconductance, same plate resistance, you have a better chance of balancing AC and DC voltages at the same time, although you will almost certainly need a trimpot to do it.

But this is all in quiescent conditions, right? As soon as you hit some signal thru the LTP, things get dynamic, and if you push the LTP into distortion all the math goes out the door.

So yes it is theoretically possible but only in quiescent condition, and then only if the OT CT is perfectly symmetric and the power tubes have identical Gm and current characteristics. I have never seen an OT with a perfectly symmetric CT, and if one exists I will guarantee it's a hi-fi OT and not an MI amp OT.

In practical terms ie what you can hear, a more or less perfectly balanced LTP, OT and power tubes system would sound pretty uninteresting to us.

[Cliff Schecht]

Yup perfectly balanced output sections cancel all even-order harmonics and will sound grainy, thin and harsh when pushed into distortion (assuming the PI somehow stays balanced). Not very interesting to the human ear.. Probably more offensive sounding, even if you play the fastest shred metal ever. Some of those guys have distortions lacking in even-order harmonics and they tend to have a "thin" sound.

[Drumslinger]

nice article, thanks robtone. pretty straight forward.

[gearhead]

Let me try a swing at things. Might be off base, but heh, lets discuss. lol.

Two things:

In a triode, voltage signal to grid does not, by itself, result in inverted voltage at the plate. Without a plate resistor, there is no voltage inversion, but there also is no voltage gain. Plate resistor does both.

Understanding the PI has been the bane of my amp understanding. In recent reading of Merlin (Christmas present!!), he explains the PI driver in terms of the cathode to grid relationship. Note this is based on my first read. He also didn't go into great detail, so had to do some interpretation.

First triode acts like typical common cathode; increasing signal swing increases voltage on grid, resulting in increased cathode current, and ultimately down swinging (inverted) plate output. Since the Rk is not bypassed, the cathode voltage does rise due to the increased current.

The other triode, whose cathode is tied to the first one, has no signal input to its's grid. So, if the 2nd triode's cathode voltage is pulled up by the 1st one, but the 2nd triodes grid voltage is unaffected by the signal, the negative voltage between the grid and the cathode actually increases. This inhibits (decreases) cathode current in the 2nd triode. Results in upward swinging (non-inverted) plate output. Which is opposite of triode #1.

Hence inversion.