LA2A Gain stage analysis

[Deepdark]

Hi there

I just began to take a look at the input and output gain stage of the LA2A. This is more like a Learning stage to me. There is some mods that consist of replacing the 12ax7 by a 12ay7 (CJ I think brought the idea) and replace the anode résistors from 220K to 120K. So, I wanted to test with spec sheet the actual 12ax7 biasing, what would be the effect of a rebiasing and do the same with a 12ay7 and 5751. I wanted to know what would be the effect if we just swap those tube without modifying the resistors (where would sit the biasing according to the cathode resistor vs load line) and a rebiasing them. I join 3 graphs (12ax7, 12ay7 and 5751) with some different biasing:

12AX7: The actual biasing, with a cathode of 1K5 and anode of 220K. I = 0.36ma. Anode voltage is about 115v. We got a biasing of about -1.5v. It place the bias point not much symmetrically so will have more swing headroom on one side than the other, right? Taking another bias point at 220K, say -1v would place the biasing in a better position, about 0.5ma and a cathode resistor of about 2K. Then, looks what happen with a load of 120K. If I keep the same plate voltage, 115v, it places me about -1.2v bias point and 0.65ma, with a cathode resistor of about 2k.  Move a little the bias point, say, -1v,and we drop anode voltage to 110v and cathode resistor go back to about 1K3-1K5. Looks good on the graph, we doubled current, bu since we're below 1ma, their is not much to worry about. From now, is my analysis looks ok?

5751 : If I swap the 12ax7 for a 5751, without any modification, where would sit the actual biasing on that one? Voltage at the anode drop to about 80-82v, plate current would be about 0.55-0.57ma and bias at about -1v. Next, if I want to get my anode voltage back to 115v, I would need to replace the cathode resistor for about 3k and a bias point at -1.5v. This place the biasing a little more equal on the load line and would swing more symmetrically. Next, if the load resistor drop to 120K, and we keep a bias of -1.5v, we get a anode voltage of about 125v, plate current of about 0.78ma and a cathode resistor of 2K. Then, if I want to get my anode voltage back to 115v, bias would be about -1.25v at 0.85ma, cathode resistor would be 1k5. Sonicaly, any advantage between a 5751 over a 12ax7?

12AY7: Again, if we only swap tubes, it would place the tube biasing at -1v, for aplate resistor of 220k, cathode resistor 1K5 and my anode voltage would be about 50v at 0.72ma?? Quiet low, isn't it? Not much place to swing, too. Go back to  somewhere around 105v-115v anode, biasing is about -2.5v at 0.5ma at 5K cathode resistor. Quiet high for a cathode resistor. Drop a 120K load, at -2v biasing, I stay in the 105v at 0.9ma anode zone, change the cathode resistor for a 2K2 and have place to swing. If i keep the 1K5 cathode resistor, then my biasing would be about -1.5v at over 1ma, but remain assymetrical and wouldn't have much place to swing one side vs the other. -2.5v biasing would even be better I think.

For now, any comments and useful things I should taking consideration? For what we see, the second stage of input amplification is basically nearly identical, exept that it's plate feed the first stage of output amplification and grid resistor is strap to the second stage plate of the 12bh7.

I don't work yet on the negative feedback line, I need to do some research into my books. But if I reduce the feedback resistor, this will result in a little less gain and more distortion?? Is the fact I change the cathode resistor would influence the negative feedback?

Next, we saw that grid résistors are all 470K to ground, exept R15, which is tied to one plate of the output tube. Why? To load the plate? The output configuration is a little harder to me. For what we see, the output looks to be cascaded (is it the right term??) so cathode of the first triode feed the second plate and the second cathode is strap to ground. So, if I want to draw my load line and Watch the biasing, I guess my cathode will be 1K + 1K (because they are in series) ?? Do we have to take into consideration the internal plate resistance?

Sorry for the long thread. Thanks for the reading and thanks for the help. I will try to make some test in the futur and will try keep this thread alive with some results.

 

[Deepdark]

12ay7

 

[Deepdark]

5751

 

[moamps]

.....Next, we saw that grid résistors are all 470K to ground, exept R15, which is tied to one plate of the output tube. Why? To load the plate? The output configuration is a little harder to me. For what we see, the output looks to be cascaded (is it the right term??) so cathode of the first triode feed the second plate and the second cathode is strap to ground. ...

It is White cathode follower.
http://www.tubecad.com/october99/page4.html

 

[Deepdark]

Thanks I'll read about it and work out the output stage then

 

[Deepdark]

Ok, I give it a look and it's much more complicated than a simple cathode follower  ;D

I'm not sure how to handle it. For what we know, Plate resistor is crucial for a good voltage swing, according to our load. if we take a closer look at the topologye: the upper portion of the tube is fed at 220v at the plate and about 120v at the upper grid. Looking at the bottom portion, plate is fed at 100v. From there, how do we handle to set biasing point? I guess we have to estimate ra and sum them because both they are in series, to which we had Rk? How do we set the plate resistor?

Thanks guys

 

[PRR]

Read the TubeCAD article on the WCF.

FWIW: the stock LA2's value for top resistor is not optimal. Broskie gives formula for a better value. "Better" for what, THD at low level? THD spectrum at high level? Maximum output for THD <5%? Best power supply efficiency? Being cheap? However the stock value is perhaps 5X bigger than it needs to be, which causes a non-negligible reduction in output ability. OTOH this is the butt-end of a *limiter*, the output is uh limited, so maybe it is good-enough.

 

[PRR]

> bias point not much symmetrically so will have more swing headroom on one side than the other

In the first stage, who cares? It directly drives a second stage. By the same thinking, this second stage will bias around 1V-2V, will never want more than 1V-2V of input signal to slam. So the first stage does NOT have to be designed for large symmetrical output.

This may change if there is loss (Volume, EQ) between stages (guitar amps); but not here.

First stage designed for good gain with wide bandwidth.

The second stage may have to be designed for large output. Here it drives a unity-gain stage to a 4:1(?) OT. Say the output could be 7V rms. Then the OT primary needs 28V rms. The WCF input is very high impedance, we can use the "680K load" part of the factory suggestions table in your first PDF. We can make 38Vrms with 300V supply, or 27V with 250V supply. This is at 5% THD (without NFB). The 38V case delivering 28V will be about 5%*28/38= 4% THD (before NFB). This may fail FCC proof-test. But we have a gain of like 50*50 or 2,500, we need a gain of what, 250? So 10:1 excess gain to NFB, THD goes down factor of 10, then 0.4% which is clean-enough for 1960.

This is meatball design. You throw some meatballs on bread and call it dinner. If the LA2 guys had wanted a *GOOD* no-THD design they would have built a 2-6V6 output stage with 12AX7 driver, plus the 12AX7 gain stages before that. "10 Watt" output stages would deliver 1 Watt or +30dBm with <1% THD before any NFB, and would "never" make rude sounds even when overdriven 6dB-8dB past where the broadcast transmitter was clipping constantly. The big GE limiters are like this. But the LA2 gang saw a path to a lower-cost limiter (the LDR) and slung some meatball tube stages at it. With growth in the radio business at this time (teenagers with transistor radios), sales were good.

 

[CJ]

you could put a mallard duck in between a UTC A-10 and A-24 and it would sound like an eagle.

Pultec?

you could put pound of hamburger in between a Peerless HS-29 and S-217-D and it would spit out a steak sandwich,

 

[Deepdark]

you could put a mallard duck in between a UTC A-10 and A-24 and it would sound like an eagle.

Pultec?

you could put pound of hamburger in between a Peerless HS-29 and S-217-D and it would spit out a steak sandwich,

What about grilled cheese 8)

 

[Deepdark]

Thanks PRR, always a pleasure to read you. I'll read it up and work the ouytput section as well as looking at the negative feedback

 

[mjrippe]

you could put a mallard duck in between a UTC A-10 and A-24 and it would sound like an eagle.

Pultec?

you could put pound of hamburger in between a Peerless HS-29 and S-217-D and it would spit out a steak sandwich,

What about grilled cheese 8)

You guys are making me hungry!  Though PRR's meatball sandwich sounds pretty good...

 

[Deepdark]

I feel more for some Pizza right now  ;D

 

[abbey road d enfer]

12AX7: The actual biasing, with a cathode of 1K5 and anode of 220K. I = 0.36ma. Anode voltage is about 115v. We got a biasing of about -1.5v. It place the bias point not much symmetrically so will have more swing headroom on one side than the other, right?

No.
The positive excursion is governed by the voltage divider between the plate resistor and the load, but the negative excursion is limited by the tube's capacity to draw current, which is always superior in normal operating conditions.
When headroom is a concern (and consequently THD at high level), the bias point should be lower than 1/2 B+; by how much is not answered without detailed analysis of the circuit and its environment.
The rule of thumb that "anode voltage should be half-way to B+" is oversimplification. It is acceptable for input stages because they generally carry signals that are much below clipping.