Grid Circuit Impedance

[physics]

Hi!

In TUT2, figure 2-62 we're shown a "typical bias supply":

   

It's stated that the impedance of the supply is about 1k || 15k at balance, or about 1k. I need to work out the supply impedance for the bias supply I'm building in an amp, so I want to make sure that I understand how this calculation was done in the book.

First, is the diode assumed to be a short? Otherwise the 1k resistor is isolated. Is this because in thermal runaway, (conventional) current would flow out of the grid into the bias network, so from that perspective the diode would be forward-biased if there is enough current flowing to develop a diode-drop's worth of voltage across it?

Second, how come the resistance of the centertapped pot is ignored? At balance, the current from the grid has to flow through 5k to the centertap and wiper, then out to the 1k and 15k assuming the diode is forward biased, so the supply impedance would be (1k || 15k) + 5k ~= 6k. Admittedly the answer isn't that different, so maybe that was why it wasn't accounted for?

Also, how come the 70u capacitor doesn't factor in? Elsewhere in the TUTs, it's stated that if we have to increase the resistances to get adequate bias voltages, we can drop caps from the bias wipers to ground to get a good AC ground for the grids. Wouldn't the 70u cap act similarly, meaning that the supply impedance is just half of the center-tapped pot at most, so about 5k?

Sorry if it seems like I'm being nitpicky, I just want to make sure I understand the thought process behind the calculation so I have a better idea of how to go about the calculations for my own projects.

Thanks!

[K O'Connor]

Hi Guys

Yes, half of the bias balance pot value should be a portion of the bias circuit impedance.

Since we are considering the amplifier in its operating state, the first approximation has the diode shorted to get a rough DCR equivalent for the supply resistance.

Grid circuit resistance limits listed on the tube data sheet is often ignored, as we see when we look at OEM products. Someone in the past decided to use a certain value and everyone has parroted it ever since, generally because of ignorance - not knowing how the value is chosen; not wanting to risk losing some "magic".

A good and bad example is the SVT power amp, which TUT3 states is an inherently unstable circuit. One of the bandaids to enforce stability is the use of a high-value grid-stop for each output tube. These are 50k, which is the maximum grid-circuit resistance listed for 6550. All three grid-stops are driven by the low-z output of a cathode follower. The grid circuit resistance for each tube is then roughly 50k. Normally, at least 90% of the grid circuit resistance would be the grid-leak resistor, but not here. One can see that the data sheet has been respected although in an unusual manner.

In some silver face era Twin Reverb and Dual Showman versions, the usual 220k grid-leak for the parallel output tubes is reduced to 68k. Neither value reflects the fixed-bias limit of 100k for a single tube.

During normal operation with tubes that have a good vacuum, grid current is essentially zero. If a tube draws grid current due to some internal failure, having a lower grid resistance likely won't make a difference to the tube's demise. However, since the bias voltage of the output tube is the most important voltage in the chassis it seems appropriate to attempt to design-in reliability wherever we can.