Single Ended EL84 tube amp plate voltage thoughts...

[Mike Sulzer]

To expand on that a bit (to clarify – I hope ):

Initially, there is a DC conventional current flowing down through the tube and down through the cathode resistor to ground.

After some (positive) charge has been hypothetically added to the top (cathode end) of the bypass cap, the cathode voltage becomes more positive by Vr.
This increase in voltage at the cathode increases current through the cathode resistor to ground, and at the same time reduces the current coming down through the tube (since Vgk has become more negative).

This is the same thing as imposing additional currents, ‘down’ through the resistor and ‘up’ through the tube, added to the quiescent current.

One approach is just to trust the model, as embodied in that equation we have been using. The "further explanation" that I like starts by saying that if we talk about an impedance into the cathode, we are thinking of current pointing into the cathode. From this point of view, the quiescent cathode current has a negative value. Applying a more positive voltage on the cathode reduces the magnitude of the cathode current, as you say. This is a change of current into the cathode in the positive direction, in agreement with the model, which says that we have a positive input impedance. Thus both the resistor and the tube discharge the capacitor.

[Chuck H]

Ah! Ok... It all clicked now. Much more intuitive and obvious than I thought. I don't know how I didn't see it.?. Like I was looking in a mirror and trying to figure out why my image raises it's left hand when I raise my right, sort of. Thanks gents.

[kevtronic]

Honestly if you want a good functioning single ended EL84 amp with only a volume control, look at the Epiphone Valve Jr schematic...

Great little amp. I actually got one with the cab for 50 bucks one and like an idiot I turned around and sold it for 150 bucks. Anyways. Ya, Epiphone valve Jr, cool amp. But, this particular amp is a diy scratch build off the chassis and tranny's from an old newcomb class A tube record player. I build them, sell/trade them, build another

[kevtronic]

I'm still here, been checking the thread on and off. I'm just trying to chew on things for a bit to see if I can't upgrade my understanding and knowledge on the subject some. Still going thru the one Aiken amp article and a couple off shoots it lead too. Get side tracked here and there, live life some, then come back to it. On a cool note I've got an old Sunn 12" speaker donated to me. I have to replace the dust cap and tinsel wire, but it functions well, so - cool deal...

[kevtronic]

Ok, this may come across as a dumb question, but being I am trying to use load lines to set the bias (something I've never really done) of my 12AX7 tube I need to ask a question to insure that I am measuring the supply voltage correctly (Vpp).

I have my EL84 output tube biased where it's staying with a plate to cathode voltage of about 255 and biased right at 100%.

Do I measure the supply voltage for the 12AX7 with or with out the plate load resistors in stalled?
If I measure Vpp without the plate load resistors connected I get 290Vdc.
If I use the 290Vdc as the Vpp with a 100K plate load resistor and construct a load line off that, I appear to do very well with a 1.5K cathode resistor, a Vg of -1.5, and an Ia of 1mA.
At least, with my present understanding of things I think I do

However, if I start out with a the plate load resistors installed I get a different Vpp and I'm playing havoc trying to find a sweet spot, like I'm chasing my own tail - lol...
It doesn't seem to matter what combination of 1K or 1.5K cathode resistors and 68K or 100K plate load resistors I use, I can't get the actual amp readings to line of with the load line.

I've been using the tube calculator that amp books provides: https://www.ampbooks.com/mobile/vacuum-tubes/12AX7/

I get the whole if your tubes aren't shutting off or burning up and you like the way it sounds, cool, run with it.
But, I really am trying learn and apply load lines, hone my skills more, take my knowledge to the next level, and apply it to this amp...

[eschertron]

However, if I start out with a the plate load resistors installed I get a different Vpp...

Are you saying that the Vpp drops substantially when you put the 12AX7 into its socket, or is this something still theoretical/modeled?

[kevtronic]

Are you saying that the Vpp drops substantially when you put the 12AX7 into its socket, or is this something still theoretical/modeled?

No, not theoretical.

I was trying to do it install the plate load resistors first (and of course that means choosing a cathode resistor too so the tube is set up some what) and take the supply voltage readings, but it never seemed to work our right in the real world after choosing a cathode resistor. So, I got the bright idea of hey, if it was 'theoretical' I'd be like, ok I'm going to set my power rail up so that it supplies 300Vdc Vpp and use a 100K plate resistor (all I got in my parts drawer, say). Which, If I disconnect everything thing from the supply/filter cap and take the measurement I figure that's pretty much it (but, am I right or wrong - well, I'm too green to say really)...

If I start with just the supply voltage with nothing connected to it (tube not in play) I get 290vdc. I can choose the plate load resistor I plan on using (100K) and construct a load line based on it. And, if from that load line I can choose a Vg of -1.5, and a Ia of 1mA bias point and that gives 1.5K cathode resistor. When I build it and take measurements they match the load line, etc...

However, if I choose to install the chosen 100K plate resistors first (set the tube up educated guess thing) and then take the Vpp supply voltage and do a load line I can't get the real world measurements to match up, per say, to the load line. When I install a cathode resistor based on doing this way the real world measurements don't match the load line. (this involves removing the educated guess cathode with the one calculated based on the load line).

Being I'm kind of a green horn at the whole load line, picking a bias point, calculating a cathode resistor to set the chosen bias point, I'm wondering when I should be taking the voltage supply reading (before or after installing the chosen plate load resistor) to create a load line based on supply voltage and plate load resistor (I'm sure it's a whole lot easier than my fried brain is trying to explain it - lol)...

You guys are great. I sure hope I don't ware out your patience

[Malcolm Irving]

When you are setting the tube up based on educated guesses are you remembering to include the 1M grid leak resistor from grid to ground? If you miss that resistor out, the grid will float and the tube will draw a lot more current, which will mess up the readings.

[Mike Sulzer]

Plate supplies for preamp tubes are soft because of the dropping R(s) used in conjunction with the bypass electrolytic(s) to get rid of ripple. The plate supply that matters is when the tube is in; there is no need to make any measurements with anything disconnected, just measure everything you can. You can measure the current through the 12AX7 by measuring the voltage across either or both the cathode and plate resistors. Make sure that they agree; then you know your measurements are good. Make sure that you have very close to zero volts (wrt gnd) on the grid. As you change the cathode resistor, and thus the current, the plate supply will change a bit. Measure plate to ground, supply to ground, and make sure those measurements are also consistent with the other measurements.

No, not theoretical.

I was trying to do it install the plate load resistors first (and of course that means choosing a cathode resistor too so the tube is set up some what) and take the supply voltage readings, but it never seemed to work our right in the real world after choosing a cathode resistor. So, I got the bright idea of hey, if it was 'theoretical' I'd be like, ok I'm going to set my power rail up so that it supplies 300Vdc Vpp and use a 100K plate resistor (all I got in my parts drawer, say). Which, If I disconnect everything thing from the supply/filter cap and take the measurement I figure that's pretty much it (but, am I right or wrong - well, I'm too green to say really)...

If I start with just the supply voltage with nothing connected to it (tube not in play) I get 290vdc. I can choose the plate load resistor I plan on using (100K) and construct a load line based on it. And, if from that load line I can choose a Vg of -1.5, and a Ia of 1mA bias point and that gives 1.5K cathode resistor. When I build it and take measurements they match the load line, etc...

However, if I choose to install the chosen 100K plate resistors first (set the tube up educated guess thing) and then take the Vpp supply voltage and do a load line I can't get the real world measurements to match up, per say, to the load line. When I install a cathode resistor based on doing this way the real world measurements don't match the load line. (this involves removing the educated guess cathode with the one calculated based on the load line).

Being I'm kind of a green horn at the whole load line, picking a bias point, calculating a cathode resistor to set the chosen bias point, I'm wondering when I should be taking the voltage supply reading (before or after installing the chosen plate load resistor) to create a load line based on supply voltage and plate load resistor (I'm sure it's a whole lot easier than my fried brain is trying to explain it - lol)...

You guys are great. I sure hope I don't ware out your patience

[kevtronic]

When you are setting the tube up based on educated guesses are you remembering to include the 1M grid leak resistor from grid to ground? If you miss that resistor out, the grid will float and the tube will draw a lot more current, which will mess up the readings.

Thanks for the reminder to double check. Not this time around, but I've done been there done that and can attest a missing grid leak will mess with things

[kevtronic]

...You can measure the current through the 12AX7 by measuring the voltage across either or both the cathode and plate resistors. Make sure that they agree; then you know your measurements are good...

Thanks for the info. The only thing you've mentioned I wasn't originally doing is calculating current across both the cathode resistor and the plate resistor and then comparing them. Actually, I've never done this before, but I will the next time I make it to the bench. Thanks for sharing, it's appreciated...

*** Edit Starts ***

I emailed a well know person in the tube amp field and here is the question and answer:

> Hello (edited out for privacy),
>
> I'm hoping you can answer a question for me, that I just can't seem to get the info I need to trust what I am doing. I've got so I understand the construction and use of load lines pretty good, short of one problem - I am unsure of were to measure the Vpp (supply voltage) in an actual amp to construct the load line. To hopefully be a little more clear, lets say I'm doing a load line for a triode preamp tube using 100K plate load resistors. Do I measure the power supply side of the plate load resistor between the plate load resistor and it's supply/filter cap (everything in place) or do I need to measure the supply voltage at the supply/filter cap node without the plate load resistors, etc installed (basically no triode tube connected)?
>
> I ask as one I am unsure and I want to do it correctly, but also because I've done it both ways with the following results. When I construct a load line with everything hooked up and in place, pick a bias point, then tweak the cathode resistor to match the load line bias point I chose, I can't seem to get my real world measurements to match the load line when I'm done. If I measure Vpp without anything connected to the supply voltage (no plate load resistors, etc) and construct a load line, pick my bias point, then build/hook things up, my real world measurements match what the load line provided nicely.
>
> So, I sure hope you can do me a BIG favor and un-boggle my brains here. I've tried reading to find the answer, but I can't find anywhere this is addressed specifically, and well, I can't (or hope I'm not) the only one wondering/confused about it...
>
> Hope this find you well,
> I enjoy the articles on your website,
> (edited out for privacy)
>
>

Hello, (edited out for privacy),

You measure the plate voltage at the supply node at the top of the resistor feeding the stage with everything connected. That voltage is the real-world voltage where the tube is running.

If you are getting incorrect results, it is likely because your tube does not match the published plate curves, which are just an average approximation, since all tubes are different, and different manufacturers have different plate curves.

Thanks,

(edited out for privacy)

And, that's something the greenhorn never thought twice about

*** Edit Ends ***

[eschertron]

I see now where you're coming from. I, too had scales on my eyes initially. if the preamp tube is all by itself on a filter node, and that filter resistor is pretty big (27k..68k or more) then the node will drop considerably when the tube is running, even with only 2ma or so current.

edit: Duncan Amps' PSUD II is a very handy tool to have, but even without you can guesstimate the Vpp at the preamp node by figuring 2v drop from the node before for each 1k of dropping resistor, for each 12AX7 on that node.

Even without knowing hard figures, a 100k plate resistor and a 1.5k cathode resistor are good 'vanilla' values that result in near-center bias for any given Vpp. Maybe reasonable given Vpp, from 150vdc to 350vdc?

[Chuck H]

Even without knowing hard figures, a 100k plate resistor and a 1.5k cathode resistor are good 'vanilla' values that result in near-center bias for any given Vpp. Maybe reasonable given Vpp, from 150vdc to 350vdc?

Hmmm... IMHE the 1.5k typical cathode resistance for 12ax7 triode stages in guitar preamps is still biased a bit cold at typical voltages. At, say, 150Vp I think a cathode R at or below the lowest typical value of 820R would be necessary to achieve center bias. I haven't nailed it down on the bench because it hasn't been a priority yet. But this is my observation.

[eschertron]

Hmmm... IMHE the 1.5k typical cathode resistance for 12ax7 triode stages in guitar preamps is still biased a bit cold at typical voltages. At, say, 150Vp I think a cathode R at or below the lowest typical value of 820R would be necessary to achieve center bias. I haven't nailed it down on the bench because it hasn't been a priority yet. But this is my observation.

I won't try to justify my statement by saying the 100k/1.5k combo is 'traditional', because I've heard the stories of how designs get unthinkingly copied across the years
I agree that when the stage is loaded down and the load line rotates, it will tend to move the bias colder. Attached is a load line chart, showing what I use for analysis. Caveat: what looks good, and what sounds good may be two different things, and we all know how well the eye hears stuff. Your preference for the kind of distortion that's generated by the bias you choose may well depend largely on the specific application, e.g., the degree that headroom is reached or exceeded. There are plenty of bias options that result in distinctive, desirable results. That's why I think of the 100k/1.5k combo as 'vanilla'.

[Chuck H]

I'm definitely not arguing "vanilla". 1.5k is certainly the most common value used for this circuit and "I" use it A LOT. I just know that when I scope such a stage (all things being equal like common plate voltage and it's first stage to be clipped, etc. etc.) I'm pretty sure I've always seen it clip in cutoff a good deal prior to saturation. But I'll want to double check that on the bench, soon. And as you mentioned, there is the voltage rise across Rk with current. Just thinking out loud, but certainly the load line can be accurate for center biasing at idle and not reflect this operational characteristic. Unless current is measured while the tube is amplifying the load line will only tell us the idle current? Which, interestingly, seems to be the popular method of using load lines for component selection and apparently often was historically.