O/T Power amp pentode grid 1 negative biasing question

[frequencycentral]

I designed and built my Murder One amp about six months ago, it's only a tiny amp, and the ~80 volts B+ is powered by a MAX1044 charge pump. I've recently revisited it and biased grid 1 to -6 volts, (again using the MAX) which has given more clean headroom. There's a thread dedicated to it elsewhere on the forum: http://www.diystompboxes.com/smfforum/index.php?topic=73222.0

Here's the original schematic: http://i210.photobucket.com/albums/bb292/frequencycentral/MurderOneV2.jpg

Here's the new schematic, with the gid 1 bias mod: http://i210.photobucket.com/albums/bb292/frequencycentral/MurderOneV4.jpg

I'm about to start work on another small amp, this time using a 5902 beam pentode. The data sheet says the gird 1 voltage should be -40 volts.

I'm just trying to get my head round grid 1 biasing. I'll be using a 555 based SMPS to derive a B+ of 150 volts for the plate, a small cathode resistor/bypass cap, referenced to ground. And a grid 1 bias of -40 volts, which I could use a MAX1044 for again.

Does this sound right? Am I understanding things correctly regarding grid 1 negative biasing, or have I missed something?

[aron]

> The data sheet says the gird 1 voltage should be -40 volts.

For a fixed bias amp - I will take the datasheet's word on it. Obviously it's different for a cathode biased amp.

http://www.londonpower.com/cath-fix.htm

http://www.aikenamps.com/WhatIsBiasing.htm

[Ripthorn]

I don't know how to answer your question, Rick, but I will be very interested in the answer.  If it turns out that negative biasing is important (or at least if it is important in push/pull) then it looks like I have one more thing to do before buttoning up my submini amp.  But if it turns out to give more headroom or something, then I will do it.

[frequencycentral]

Thanks for the links Aron, very useful reading. I've 'designed' a dozen tube circuits but I know I still have a lot to learn.

So.........there's a couple of way to do this with the 5902 I mentioned:

1. Reference the grid to ground, apply 40 volts to the cathode using a resistor voltage divider, apply 150 volts to the plate.

2. Apply -40 volts to the grid, reference the cathode to ground, apply 150 volts to the plate.

Is that a correct interpretation of the second link Aron provided?

It seems to me that method #1 is easier to achieve that method #2, as it's just a case of dividing the B+ with resistors to get 40 volts, and therefore the ground referenced grid will be -40 volts with respect to the cathode.

@ Brian: This is on my mind right now due to Aldas giving me the heads up on how he biased his Murder One. I did similar and got more clean headroom straight away, as an experiment I applied -12 volts to the 5672 grid and got less clean headroom, so around -6.5 is optimal for the 5672 - and it doesn't have a cathode to bias as it's a directly heated tube, it just happened that I was aware it's easy to get -6 volts from the MAX1044 as well as using it as a charge pump doubler. The 5902 says a maximum of -55 volts. So I would say at this point that if 'Kitchen Sink' has both grid and cathode referenced to ground you should be able to get more out of it by biasing the cathode to 40 volts. Maybe time to get the breadboard out.........very interesting stuff!

[Ripthorn]

So Rick, would I apply the 40 volts to the + side or - side of the electrolytic cap on the cathode?  In other words, do I apply the 40 volts directly to the cathode, or do I reference it to ground meaning I have 40 volts where the ground connection would be?  That sounds really easy since I have some large value zeners (39V or so) that I could chain up, or I could do a voltage divider.  I have a single loose 5902, but I am wondering if there is any difference whether the tube is in SE or PP configuration.  On the topic of the Kitchen Sink, I have had to trim back some of the options as some of the less important ones didn't play too well with the more important ones.  But believe me, the tone palette still deserves the title.  I still need to put together a Murder One I think, since it is getting rave reviews and I have everything but the transformer.

[frequencycentral]

Brian, read both the links Aron provided, then read the 'Characteristics and typical operation' from the 5902 data sheet: http://www.scottbecker.net/tube/sheets/141/5/5902.pdf

See if you draw the same conclusion as I have.

.....or do I reference it to ground meaning I have 40 volts where the ground connection would be?  That sounds really easy since I have some large value zeners (39V or so) that I could chain up, or I could do a voltage divider.

I think that's the way to go. I need to do some more thinking and some experimenting to be sure though. If you feel like trying it that would be great, I doubt you would harm the tubes whatever happens. It seems that fixed biasing is less of a compromise than cathode biasing (which is how you have it set up right now). I don't think I'll find time to rig up a circuit on breadboard before August as I'm planning to finish PG this week, then I'm off to Greece and Turkey for a break. I'll be all over this idea when I get back though!

[R.G.]

Tubes are somewhat like JFETs. They are depletion mode devices, meaning that they conduct as heavily as they can unless you do something to throttle them back.

What matters to a tube is the voltage between grid 1 (the control grid) and the cathode. In triodes this is all that matters for current flow. In pentodes and power beam tubes, the G2 and G3 get into it, but they have much less effect than Q1, and are properly used for other things. The screen grid in particular for pentodes and PBTs can be thought of as a "gain control", which sets the effective transconductance. The suppressor grid and beam forming electrodes are there to force secondary emission electrons back onto the plate, preventing the low-voltage kink in characteristic that was the hallmark of four-element tubes. The "KT" in "KT88" and other tubes stands for "kinkless tetrode"; it's a power beam tube named in European, especially British, style.

So whatever you do for bias, the tube separates out the voltage between cathode and G1, and uses that as the effective bias. This can be all an external negative bias, all cathode self-bias, or a combination of the two. Screen grid is set quite a bit positive; it's for accelerating electrons to the plate, and so changing it's positive voltage with respect to the cathode determines how "hungry" the plate looks to the cathode, and therefore has a major effect on the gain of the tube. It does act as a second signal input if you do it that way, but it has a much lower "gain" to signal than G1. Don't muck with the suppressor grid in general. Hook it to the cathode or more negative.

[Ripthorn]

So if I understand you correctly, RG, if I were to hypothetically apply a +40V to my cathode, and I had a B+ of 110V, then to the screen, it looks like a 70V differential, correct?  That seems to me that the plate would then look less "hungry", or did I completely misinterpret your explanation?  I am thinking of how Rick added a negative voltage to grid and it makes sense that that would increase some headroom because you are maintaining the screen-cathode voltage differential while increasing the grid-cathode differential.  Anyway, I'm off to read more about this issue, so we'll see what crops up.  Thanks, as always, for taking the time Aron and RG.

[aron]

To tell you the truth, I would just cathode bias the tube and be done with it. It's the most simplest, easiest way to bias a tube. So many "class A" amps can't be that wrong. Check out a typical Fender Champ etc....

http://galileo.spaceports.com/~fishbake/champ/chmp.gif

Here's another cathode biased amp:

http://www.frontiernet.net/~jff/pipsqueak_pentode/Pentode_DIY_guitar_amp.html

[R.G.]

So if I understand you correctly, RG, if I were to hypothetically apply a +40V to my cathode, and I had a B+ of 110V, then to the screen, it looks like a 70V differential, correct?  That seems to me that the plate would then look less "hungry", or did I completely misinterpret your explanation?  I am thinking of how Rick added a negative voltage to grid and it makes sense that that would increase some headroom because you are maintaining the screen-cathode voltage differential while increasing the grid-cathode differential.  Anyway, I'm off to read more about this issue, so we'll see what crops up.  Thanks, as always, for taking the time Aron and RG.

]
Let me go at it from a different direction.

A tube is a fairly dumb device. It only knows what touches it, and has no clue what the voltages and currents may be somewhere else. In particular, the grids only have effects based on their voltages to the other parts of the tube. The plate and cathode handle all of the current in the tube (* there is a minor exception, but this is the way to start thinking about it). Ideally, the grids would conduct NO current, only supply voltages to affect the electrons on their way between cathode and plate. A tube knows about grid to cathode and grid to plate voltage. What you do with your power supply makes no difference whatsoever to the tube, except as it affects the interelectrode voltages.

G1 is the nominal control grid. It sits very close to the cathode and is more negative than the cathode. It works by repelling negatively charged electrons back into the space charge area, the cloud of electrons floating over the cathode. The more negative it is with respect to the cathode, the more it keeps from getting by and the less current that flows to the plate. When it goes to 0V with respect to the cathode, the electrons don't know it's there and fly freely to the plate.

The screen grid "screens" the plate from the effect of the control grid and also accelerates electrons toward the plate. It has to be positive with respect to the cathode to pull the electrons toward the plate. Most of the electrons miss the screen grid mechanically, as it's quite open. Some small fraction hit the screen grid, and this accounts for the only other current in there. The harder the screen grid pulls on electrons - that is, the more positive it is with respect to the cathode - the faster the electrons move to the plate, and the more electrons move to the plate, after being let through by the control grid. In effect, the higher the voltage on the screen grid, the higher the gain of the tube. There is one design for a voltage controlled amplifier using the screen grid of a pentode for controlling gain. But again, it only matters where it is with respect to the cathode. It matters a little where it sits with repect to the plate; if you make it more positive than the plate, screen grid current skyrockets as the electrons would prefer to be over at that wickedly positive screen grid than the dull old less positive plate.

The suppressor grid only exists for one reason: to suppress secondary emission from the plate. Electrons are flying toward the plate, get near the negatively charged suppressor grid, and may slow down, but they get through. With no suppressor grid, they would be flying so fast that a fraction would bounce right back off the plate. The suppressor grid both slows the incoming and forces the slowed ones who do bounce back onto the plate.

So it's only the differential voltages between plate elements that matters. If you have a 110Vdc supply and eat up 40 of it in a cathode resistor, well you only have 70V left between cathode and plate.  Pentodes may not work all that well on such a low voltage. I know big ones don't.  If you have 110Vdc between cathode and plate and force the control grid further negative some way, you now have a much larger power supply to produce output instead of producing heat and a bias voltage like they would if you used cathode resistors.

Cathode biased amps are not necessarily class A. They can be class AB as well, or may use combined cathode and fixed biasing.

[Ripthorn]

Thanks for the clarification, RG.  I think what you said confirms what I was thinking, though I am not sure I expressed it too well above.  Gives me some good stuff to think about.

[aron]

> Cathode biased amps are not necessarily class A.

Yes, I still really quite don't understand  - well maybe I do, but in any case, that's why the quotes were there. Thanks!

[frequencycentral]

Thanks for the excellent explanation R.G.

So it's only the differential voltages between plate elements that matters. If you have a 110Vdc supply and eat up 40 of it in a cathode resistor, well you only have 70V left between cathode and plate.  Pentodes may not work all that well on such a low voltage. I know big ones don't.  If you have 110Vdc between cathode and plate and force the control grid further negative some way, you now have a much larger power supply to produce output instead of producing heat and a bias voltage like they would if you used cathode resistors.

I think this confirms what I've been mulling over. I'm still getting to grips with it though. Are the absolute maximums for plate voltages on data sheets quoted with respect to the cathode? If so, using the 5902 as an example (the plate maximum being 165 volts) from a 200 volt supply, the plate would be 160 volts with respect to the cathode, the cathode would be 40 volts and grid 1 grounded? That would give maximum available gain? Or is it better to reference the cathode to ground, apply 160 volts to the plate and apply - 40 volts to grid 1? Or are they effectively the same?

[aron]

I always thought it was respect to ground.

[R.G.]

Yes, I still really quite don't understand  - well maybe I do, but in any case, that's why the quotes were there. Thanks!

This one threw me for a while back when I was learning tubes from a zero starting point. It's back to this concept that the tube only knows what it sees on its pins. All the tube knows is that the grid is more negative than the cathode. It doesn't care whether the grid is externally held low, or the cathode elevated by its own current.

Admittedly, it is quite rare to see A class AB amp which is cathode biased, unless it's only barely nudged out of Class A and a tiny step into Class AB - like the AC30. It takes some effort and scope traces to convince people that the AC30 is class AB, but you can scope it and show that the current in one tube does go to zero for a bit when the thing is driven to clipping on the output stage. The thing that makes this rare is that the average current through the combined cathodes is what makes the bias voltage, and that current is hugely different at different drive levels as soon as you exceed Class A. In the AC30, the average current in the cathodes goes up when the thing goes into AB, and that has the effect of increasing the back-bias, making the bias colder when the thing is really cooking. That's not altogether a bad thing. But it is confusing.

The variable nature of average cathode current is probably why most designers bag the cathode resistors and just go for fixed bias. Fixed bias lets you get more power claimable in advertising for the exact same power supply, which is a good thing if you're an MBA. 

Are the absolute maximums for plate voltages on data sheets quoted with respect to the cathode?

Yes, although that "absolute maximum plate voltage" number contains more than just voltage considerations. It will probably have some nod to powre dissipation as well, as well as arc-over potential. Remember that in a push-pull stage, the two plates are connected by a transformer. When the active plate pulls down from the B+ supply to 50V (which is a realistic situation under max drive) the other plate is forced up by( B+ -50V) to nearly twice the power supply voltage. In a pentode, you're usually not so worried about plate arc-overs to the grids as you are plate to suppressor grid (which is often connected to the cathode internally by a wire) or from post to post in the tube base. There are tricks you can play to get plate voltages more than the recommended ones, but in general if you simply make your B+ be the recommended max, you're OK.

If so, using the 5902 as an example (the plate maximum being 165 volts) from a 200 volt supply, the plate would be 160 volts with respect to the cathode, the cathode would be 40 volts and grid 1 grounded?

I would not hesitate to do that. But if I was running it in AB with fixed bias, I'd only use 165V for the plate supply. Also, if you're running it from 200V and putting 40V across the cathode resistors, you're usually fairly deep into class A. Check the power dissipation of the tube. Multiply the cathode current times the plate-cathode voltage (which crudely compensates for screen current) and see what the power is. How does that compare to the absolute maximum power rating? The tube WILL at some point become an incandescent light bulb for a few milliseconds if you exceed the power rating long term.

That would give maximum available gain?

Does not compute. Gain is dependent on the external circuit, especially the screen grid voltage, even at the same interelectrode voltages.

Or is it better to reference the cathode to ground, apply 160 volts to the plate and apply - 40 volts to grid 1? Or are they effectively the same?

I would think of them as the same. They're not of course. Cathode biased you're also running some serious power in those cathode resistors, and buying a bigger (voltage) and stronger (current) power supply to do it. But the tube operation is to a first order the same.

Which brings up the cathode resistor bypass capacitor. It is critical that this resistor is big enough to more than pass any low frequencies you want to pass, and that it's high quality enough and rated for enough ripple current not to burn up (i.e. boil it's electrolyte), and thermally rated to live right next door to those hot cathode resistors.

I personally don't like cathode biased output stages, but it is a personal opinion. In a small to tiny amp, it may be perfectly fine. Finding out whether it is or not is of course what the design process is all about.

[maarten]

Hello,

I have been using the 5902 to my satisfaction with cathode resistors varying from 270 to 390 ohms, bypassed by a cap (4,7 uF). I do not doubt that this gives me the full
1 Watt that this tube is capable of (approximately, did not measure it but comparing it to other small amps, e.g. with a TBA 820). Using either 5672's or 6088's in the preamp you need three stages to get there. Connect the resistor to the suppressorgrid to ground by using a cap (I used 1 uF, but you may have different taste) for good amplification. Together with a good speaker this gives you still too much noise for bedroom practice (according to my family), so you might want to settle for a relatively inefficient speaker....

Maarten

[frequencycentral]

R.G. - thanks once again for a great explanation.

So can anyone suggest any amp designs I can look at in which grid 1 is fixed biased with a negative voltage? All the designs I'm aware of appear to be cathod biased.

Thanks !

Rick

[Lurco]

almost every Marshall amp at schematic heaven?

[maarten]

Rick,
I have been looking for designs for the 5902 of course, but I can not remember having seen any schematic with fixed negative biasing.  I do remember a post however, by someone who had been experimenting with this and who had discovered that the full - 55 volts were needed for the circuit to work well. This is what made me stick with cathode biasing, so I just can use about any type of easy available power transformer in the range of 30+ volts (for SE types of amps).
Maarten

[Ripthorn]

Rick, you should ask Dana (UR12) over at ampgarage as he has built basically a submini trainwreck express.  the actual express is fixed bias, and his submini design may well be too.  Check it out.