Split Load vs Voltage Divider

[surfsup]

Curious as to your opinions on the differences of these two circuits (values are for illustration purposes).

[tubeswell]

The 'voltage divider' has a lower AC load than the 'split load' in that schematic - all other things (i.e.; tube types and other parts values etc) being equal. Couldn't work out what the actual load lines are without seeing all the circuit information for the driver stage.

[Tillydog]

Interesting question. The Split load will have a lower ouput impedance. I put them both through LTSpice. The gain is about the same, but the frequency response is markedly different. The drop at high frequencies in the case of the potential divider is expected because of the combination of the miller capacitance of V2 and what is effectively a high value grid stopper.

I can't explain the low end differences, though. Go on, spot my mistake!

Andy

[martin manning]

Replace the 220k to ground in the split-load case with a 1M and the low end should look similar. You have a 0.1u and 220k RC there vs a 0.1u and 940k high-pass.

[Tillydog]

Give that man a ceegarrr!

[roberto]

Yes, with that 220k it's a 7,2 Hz high-pass filter.
You won't ear the low end attenuation, but the lower impedance will change the feel when the previous stage is driven hard.

[surfsup]

whoops, sorry that 220k should have been a 470k...! Meant to have the grid load resistors the same.

[roberto]

You need the same load seen from the cap, not the grid leak.
So 940k (1M it's ok), not 470k.

[surfsup]

ding...got it. they are in series between the cap to ground.

[gingertube]

The only real difference is usually not of concern.
With the split anode load SIGNIFICANTLY more of any residual power supply noise on the B+ will end up at the signal output.

Interstage attenuator:
1st - the rail noise is divided down, bottom of the divider is the impedance looking "down" into the tube anode working against the anode load resistor.
Then
2nd the noise at the anode is then divided again along with the signal in the interstage attenuator.

Split Anode Load:
1st - the divider is much smaller, more of the B+ noise will appear at the junction of the split anode load. (Smaller B+ to output resistance and the rest of the anode load adds to the impedance looking into the anode).
2nd - interstage attenuator doesn't exist in this scheme so no additional division.

With a well filtered B+ to this stage you probably won't notice any difference.

Unless you really need to lower output impedance offered by the split "tapped" anode load then use the interstage attenuator for preference COZ it will give superior noise performance (whether you can notice it or not)

Cheers,
Ian

[Reeltarded]

Maybe drinking koolaid, but recently I found some real magic in varied split loads on my 4/5 stage amp. Splitting at around 70% signal but leaving the mixers 470/470 in the third stage really gave this thing a crying fundamental that I didn't find in that stage by balancing acts with the mixers.

Prolly just lucky.

[martin manning]

Unless you really need to lower output impedance offered by the split "tapped" anode load then use the interstage attenuator for preference COZ it will give superior noise performance (whether you can notice it or not)

The output impedance is not necessarily much different either... In the case here with a 100k plate load vs. a 50k-50k split and 63k ra, the difference is only 10%.

[surfsup]

In using TINA, i can't seem to get the software to model the HF dropoff any differently, regarding physical proximity to the tube. Any resistor in series with the grid affects HF. So i assume it assumes the resistor is at the pin of the tube. Question, possibly a stupid one considering I never considered this before, does it affect HF rolloff less moving the physical resistor further from the tube?

[martin manning]

The software knows nothing about the physical location of the components. It only models the response of the circuit based on the electrical connectivity and properties of the signal source you have specified. The HF roll-off of the audio signal is probably what you should be most concerned about, and it is giving you good information about that. The reason that the grid stopper is best located at the tube pin is so that the majority of any stray RF will enter up-sream of the low-pass filter created by the resistor and the Miller capacitance, and therefore be effectively suppressed. That "signal," or any other real-world effect, does not exist in the simulation unless you include it explicitly.

[Tillydog]

Another ceeegarrr here, if you please....

The only real difference is usually not of concern.
With the split anode load SIGNIFICANTLY more of any residual power supply noise on the B+ will end up at the signal output.

And how! I would not have appreciated this - thanks.

Here's 50mV of 50Hz ripple on the supply with a 50mV 1kHz input signal (peak voltages, I think). Look at the envelope of the output.

@surfsup - the results will depend on the tube models in your sim. I'm using Steve Bench's drop down models in LTspice, and am pretty impressed with how well they work.

Andy