Hoffman Amplifiers Tube Amplifier Forum
Amp Stuff => Tube Amp Building - Tweaks - Repairs => Topic started by: Quatro on September 11, 2011, 08:31:27 pm
-
Can any one shed any light on the relationship between screen voltages and plate voltages in a power tubes (say 6V6)? How does altering the voltage difference between these two structures affect tone? How does it affect power? How does it affect the safe or optimal operation of the tube? Any guidelines or explanation of principles would be greatly appreciated.
Thanks
-
How does altering the voltage difference between these two structures affect tone?
It doesn't, at least not directly.
How does it affect power?
It does however, affect this.
Do you have a solid grasp of how triodes work? Do you understand how you can change the grid voltage to alter plate current, but that you can also change plate voltage to alter plate current? The amplification factor of a tube is a measure of how much more effective the grid voltage change is than plate voltage change at changing plate current. The grid is more effective because it is physically closer to the cathode than the plate.
Take a triode, and slap an additional grid between the existing grid and the plate. You now have a screen grid (G2). It is closer to the cathode than the plate, but no where near as close as the control grid (G1).
Have you looked at the plate characteristic curves of triodes and pentodes (or beam power tubes, like the 6V6)? Triode curves generally slope up and to the right, while pentode and beam tubes have curves which are mostly horizontal. Think about what this really means. The horizontal axis of the graph is plate voltage and the vertical axis is plate current.
With a triode, as we increase the plate voltage (move to the right) with a given grid voltage (one of the curves), the plate current rises. Some grid lines rise more quickly that others, but what this really means is we can envision the tube as having an internal resistance which is relatively small, so that more voltage pushes more current through the internal plate resistance.
With a pentode, the gridlines (above some amount of plate voltage) are nearly horizontal. This is more strictly true with most beam power tubes (like 6L6, 6V6, KT66, etc) rather than true pentodes (EL34, EL84, etc). This means that the plate current is largely unchanged by even a large increase in plate voltage. If you took a 20M resistor and applied 20v, you get 0.001mA of current to flow through the resistor; apply 200v and you get 0.01mA, or just 0.09mA more. Pentodes don't have an infinite internal plate resistance, but it is very high (like several megohms). Ultimately, this means that above some critical value, the plate voltage has very little impact on plate current.
Note that most data sheets show pentode plate characteristics at a single set screen voltage.
Some datasheets have triode curves for pentodes and beam power tubes. The curves look an awful lot like typical triode curves. Triode operation is when the screen is connected to the plate. Now, plate current rises in a typical manner when plate/screen voltage rises. What does this mean? The the voltage on the screen of these tubes impacts the plate current the way that plate voltage impacts plate current in a triode.
Most data sheets show plate curves for only a single screen voltage. Here's a rare exception for the 6L6GC (http://www.mif.pg.gda.pl/homepages/frank/sheets/093/6/6L6GC.pdf). Pick any plate voltage for the 0v gridline on the top graph on page 4, and compare it to the same plate voltage on the 0v gridline for the top graph of page 7. These two graphs show characteristics for a screen voltage of 250v and 400v, respectively. The higher screen voltage always results in a higher plate current for a given grid voltage.
You can also see this effect, for the 0v gridline only, shown in the top graph of page 6. That graph allows a designer to quickly see the peak plate current (occurring when the grid momentarily reaches 0v) at various screen voltages.
-
So what does this mean?
If you want the maximum output power for a given grid input signal, the screen voltage must be held rock solid. The plate voltage can and must bounce around as a result of plate current variation working against the load impedance. If the screen voltage bounced around as much as the plate, then you'd need a probably bigger input signal and still get less output power (take that part on faith right now, though I can point you to a reference). If the screen voltage sags appreciably, then the peak current that the tube can reach will be limited, in the way shown by the graph on page 6; if the voltage drops, the 0v gridline that the tube is constrained by drops to a lower plate current value.
So, in big class AB amps, where plate current drawn from the power supply varies widely, there is often a choke feeding the screen node of the power supply. The intent is to gain all the ripple reduction while reducing the d.c. resistance, and therefore the likelihood screen voltage will drop due to increased current draw.
A different perspective on this is that screen dropping resistors should be kept as small as possible while still providing some protection to the screen. Others, who want sag and not necessarily maximum power output might use very large-valued screen resistors (like several kilohms) to introduce power compression by dynamically reducing the screen voltage.
-
A couple of other quick points...
1. It should be obvious (after you grasp all of the above) that if the screen voltage drops to 0v, the tube will pass little, if any, plate current. So in a non-functioning amp, you generally want to check for the presence of roughly correct d.c. voltages first.
2. You should not worry about the screen voltage being a bit higher than the plate voltage in most cases. While the screen voltage ideally stays steady during a signal, the plate voltage bouncing widely above and below its idle value (like 200-300v in many amps). This is normal, as long as the plate voltage doesn't fall momentarily to a too-low value (often something between 40-80v, depending on tube type). So if your screen is 10v higher than the plate at idle, this is a non-issue.
3. You should know that power = voltage x current. In an output tube, the power output is in fact a plate voltage swing (in RMS volts) x a plate current swing (in RMS amperes). Assume for a second we have a tube which has an input signal applied, and that results in a plate current swing. The tube has some kind of load impedance attached, and current x impedance (or resistance) = voltage. If the current is rising and falling, the voltage drop across the load is also rising and falling. That voltage swing across the load impedance leaves less of the total supply voltage to remain across the tube, so the plate voltage is falling as plate current is rising and vice versa.
Now think back about that graph on page 6, and how the 0v gridline drops to a lower plate current as screen voltage is reduced. If the maximum current is reduced, that means the size of the plate current swing is reduced, and therefore the plate voltage swing is also reduced. Less voltage swing x less current swing = less power (in 2 ways).
-
I won't talk "tone". That's between the ear and the beer.
Plate voltage sets the maximum plate voltage swing.
G2 voltage sets the maximum plate current swing.
There is NO abstract engineering reason they need to be similar, different, or any relationship. In some (mostly non-audio) designs they are set very different in no particular way.
> How does altering the voltage difference...?
Difference affects Power Supply costs.
Plate current is large. You design the main supply for this load.
G2 current is less, but very variable.
It is generally a poor idea to "voltage-drop" G2 supply from plate supply. Resistor dividers sag when G2 current rises. This leads to high G2 voltage at idle and sagged G2 voltage at what should be full power. This can be reduced with a fatter divider, more HEAT.
It is usually inconvenient and costly to build a separate supply for G2.
Therefore the most-practical audio amp runs G2 very near full plate voltage.
G2 is more sensitive than plate for ripple. We usually add one stage of B+ filtering just for G2 (much cheaper than filtering the whole Plate+G2 supply). This filter will have some drop, and therefore some sag from idle to full roar. If the idle drop is small, 10%, it will probably be fine.
The Usual Audio Pentodes have been _designed_ to allow a happy operation with G2 very near Plate voltage. This usually means a Mu(g2) of about 10.
There are tricks.
For MAXimum peak current you want a lower Mu. You can get higher plate current with lower G2 voltage, and in particular a lower plate-drop at max current. TV sweep tubes often work Vplate 300V, Vg2 150V, and Mu near 4. This also requires a large (costly) cathode and extra heater power. Overall it is a not-the-best way to design an audio amp; except when the Tube-TV racket collapsed, sweep tubes were a glut on the market and several classic amps were built around them.
For the very maximum audio power you want a high plate supply voltage, high impedance load, low G2 voltage (and low plate-drop). 6550 has a condition for 600Vp 300Vg2 and 100W per pair. Although this seems to mean a separate G2 supply, actually there is an elegant trick which incidentally suits the old costly low-rated silicon rectifiers. Note that for tubes much smaller than 6550, the "hi-Z load" becomes "too high" for practical good performance.
Tube designers knew all this. Most audio tubes have been proportioned (max Pdiss, max Vp, max Vg2) for some reasonable range of reasonable (and competitive) conditions and output. Small variations won't have large effect. Large deviation will probably exceed some spec or just work poorly.
-
Phenomenal thread. I learned a hell of a lot!
-
It's pentode week @ Hoffman. I showed back up at the right time. I'm getting questions answered I was afraid to ask.
>there is an elegant trick which incidentally suits the old costly low-rated silicon rectifiers.
I'm guessing this would be a stacked pair of bridge rectifiers? This weekend I finally did a little experiment using a voltage doubler where I tapped the screen from the center point. The bad thang here is the >1V of ripple. (this was a cheap stuff laying around experiment).
-
Yes, I suspect that the "center tap" off the voltage doubler is only half-wave rectified.
-
stack 'em as high as you need to...
--DL
-
Yes, I suspect that the "center tap" off the voltage doubler is only half-wave rectified.
build a low volt variant and look at it with your scope. :-)
--DL
-
There is a problem with RicharD's stacked bridge rectifiers circuit:
(http://i497.photobucket.com/albums/rr333/valvetone/Dual_PSU.png)
The transformer secondary is short circuited through two diodes, one in each bridge.
Dummyload's stacked bridge circuit is fine, because the two secondaries are isolated from each other.
(http://i497.photobucket.com/albums/rr333/valvetone/Stacked-PS.png)
I'm not sure about the capacitors after the inductors though. The negative end of the plate supply capacitor probably should go to ground, rather than to the positive end of the screen supply. (http://www.guitargear.net.au/discussion/Smileys/default/head_scratch.gif)
-
The negative end of the plate supply capacitor probably should go to ground
why do you think it wouldn't work? i believe it will work either way.
--DL
-
Hmmm.... :nice1:
Brad :think1:
-
why do you think it wouldn't work? i believe it will work either way.
It should work either way, but having the capacitors stacked will allow ripple from the plate supply to be superimposed on the screen supply. This may raise the amplifier's background hum level.
-
having the capacitors stacked will allow ripple from the plate supply to be superimposed on the screen supply
my thoughts were that they are series aiding and would reduce ripple?
sim time?
-
sim time?
I'm an old luddite who doesn't cannot do sims. This would normally be a good excuse to go build something ( not that I've ever needed much of an excuse ) but I don't have any filter chokes. :sad:
Umm... Perhaps it really is sim time. :smiley:
-
There is a problem with RicharD's stacked bridge rectifiers circuit:
(http://i497.photobucket.com/albums/rr333/valvetone/Dual_PSU.png)
The transformer secondary is short circuited through two diodes, one in each bridge.
Where? Assume a moment in time, with one end of the secondary "positive", the center-tap at a mid-value, and the other end "negative". The voltages don't have to be absolute or referenced to ground, they just need to have this relative orientation to each other. Now, given your voltage levels on the secondary taps, mark which diodes would be forward-biased in that instant.
If you do this carefully, you'll see the only "short" is through diodes from the two bridges which each connect to the midpoint of the series caps; in other words, a short to a point that would be connected anyway.
>there is an elegant trick which incidentally suits the old costly low-rated silicon rectifiers.
I'm guessing this would be a stacked pair of bridge rectifiers? This weekend I finally did a little experiment using a voltage doubler where I tapped the screen from the center point. The bad thang here is the >1V of ripple. (this was a cheap stuff laying around experiment).
Why the need to stack the bridges? Yes, given your doubler issues, it would make sense to take that approach, but that is really the fault of the doubler (which accepts sub-optimal ripple performance and lower current for doubled output voltage).
Kagliostro posted a question on a amp power supply (http://www.el34world.com/Forum/index.php?topic=11999.0) back in June that demonstrated one way to do what PRR is talking about. Yes, you'd think there would be a ripple issue with using the center-tap of the PT secondary, but there isn't one (that's too bad, anyway).
The real drawback of that particular amp's supply is you only get a single filter cap filtering the screen node. So do what Marshall does (http://www.el34world.com/charts/Schematics/files/marshall/jcm800_superbass_100w_1992.pdf), and direct the CT to the junction of series caps. Marshall most likely did this because it forces voltage sharing between the caps. You now don't need bleeder resistors to enforce the voltage sharing (though they are still a good idea to drain voltage on shut-off.
The downside of the series caps is the effective capacitance of the filter is reduced.
I'm not sure about the capacitors after the inductors though. The negative end of the plate supply capacitor probably should go to ground, rather than to the positive end of the screen supply.
You could do that. But your cap now needs to be rated for the full plate voltage, rather than half-voltage like the series caps in the first filter of larger Fender amps.
-
you'll see the only "short" is through diodes from the two bridges which each connect to the midpoint of the series caps; in other words, a short to a point that would be connected anyway.
What about this path?
(http://i497.photobucket.com/albums/rr333/valvetone/Dual_PSU_short.png)
-
Interesting tread
I'm waiting for further advice
Many thanks for sharing this knowledge !
Kagliostro
-
Good catch Darryl.
It looks like Richard has the orientation of bridge D1 rotated 180 degrees. If he flips that top bridge so that it is facing the way D2 is facing, it will correct the problem. That would eliminate the need for dual secondaries.
-
sim time? My preference is research, but I haven't been able to find an answer to the question of ripple at the the half voltage portion of the doubler. I am not aware of any production amp that uses the half voltage portion -- even though this would be a convenient solution for a KT-88 that likes 600 plate volts and 300 screen volts. Street smarts suggests that something's amiss with the half voltage portion of the doubler. Maybe it IS time to experiment!
-
Awesome discussion guys! I missed that thread from June and learned a lot there too.
Keep on rockin' :guitar1
Chip
-
Jjasilli reminded me of a question -- Why do we always run screen voltages so brutally high? For example, 5881 data suggests max
voltages of 360V for the plate and 270V for screen and we think nothing of running the screen 99% of plate. Is this just the way it's done or is there a specific reason why it is necessary? Maybe I don't get it..... Jim
-
IF I am comprehending HPB's and PRR's excellent posts, the higher the screen the higher the amplification (mu). And audio pentodes are designed for the higher screen current. You can certainly lower the screen current by using larger resistors, but then you get into more sag than you want. Some builders prefer using lower output/power, so they do lower the screen voltages.
Now, please don't take my explanation as correct. :laugh: I mostly wrote this out to help myself better understand how all this works, and I'd bet good money I've got at least some of it wrong. But getting corrected is part of my learning process as well. I ain't proud. :icon_biggrin:
-
there is an elegant trick which incidentally suits the old costly low-rated silicon rectifiers.
I think Figure 4B on this page has that elegant trick:
http://sound.whsites.net/valves/design2.html#s5 (http://sound.whsites.net/valves/design2.html#s5)
edit: link fixed, thanks to member Craigar for pointing out the dead link and finding a working one
-
Thank-you VMS -- research prevails (though I couldn't find it): "The centre-tapped bridge is a very common arrangement, and no balancing resistors are needed. It provides a half voltage point, and ripple frequency is double the mains frequency on both full and half voltage points. The voltage doubler produces ripple at the mains frequency on the half voltage point." [Emphasis added]
EDIT: So now we see that the center-tapped bridge voltage doubler yields full wave rectification at the center tap. BUT, at what voltage; or more appropriately, at what percentage of voltage (or am I missing something in the article)? The point being that the same amount, or percentage, of voltage delivered to the screen(s) as to the plate(s) is unacceptable, and needs to be reduced.
EDIT: Whoops, it's half-wave rectification on the half voltage point. Full wave rectification would produce 2X the mains frequency. Also the amount of ripple voltage : VDC is unstated but probably not good.
-
see my sweep tube thread.
http://www.el34world.com/Forum/index.php?topic=12226.0 (http://www.el34world.com/Forum/index.php?topic=12226.0)
see the 5686 thread.
http://www.el34world.com/Forum/index.php?topic=12190.0 (http://www.el34world.com/Forum/index.php?topic=12190.0)
both doubler examples in the link: http://sound.westhost.com/valves/design2.html#s5 (http://sound.westhost.com/valves/design2.html#s5) were used the sweep amp used CT doubler (fig. 4B ckt on left), and richard used doubler (fig. 4B ckt on right) in potty mouth amp in 5686 thread.
in this thread i presented a stacked doubler that uses two FWB but requires two separate secondaries - be it two secondaries on the same transformer or separate transformers. essentially it is two fig. 4A bridge circuits stacked. there is debate between darryl and i about the ripple with the LC networks left stacked or individually grounded as to which is a cleaner power source.
richard presented a single CT winding with two FWB stacked - the jury is out as to weather or not that is viable. as drawn, darryl claims it presents a short ckt to the winding, HBP says it will work if the FWBs are tied (+) to (+) and (-) to (-). anyone breadboard or SIM this one yet?
hopefully PRR will chime in and save us all. :worthy1:
in the meantime, don't take any wooden diodes...
peace.
--DL
-
I can't do "Cork Sniffer" type descriptions, but
On my early 50 watt heads, I used 3.9K screen grid resistors which dropped the voltage to the screens from 477 B+ (plates) to 441 (screens)
The amps had a very creamy smooth type feel to them doing this.
(http://www.el34world.com/Hoffman/images/2034skem.gif)
-
There you go--- and I'll bet the amp runs cooler and the tubes last longer.
-
Please note my corrections in the prior post to my response to VMS's post. It is half weave rectification at the half voltage point of the voltage doubler. This would explain why the half voltage point is generally not used for the screen & preamp B+ supplies.
-
Back to the main point of this discussion. OK, so after a tipping point on the chart, which is quickly reached, changes in a pentode's plate voltage have little effect on plate current. But changes in voltage on g1 and g2 do have their effects. So, what does this tell us in terms of DC operation, and AC signal processing?
-
richard's ckt doesn't work. it shorts the secondaries. if you guys can come up with any other variations, i'd be happy to give it a go.
the two diode doubler is a full wave at the top of the stack and a 1/2 wave at midpoint. no more suspicions.
the CT diode bridge doubler is full wave at the top of the stack and at midpoint - no doubts on this one - just proofing.
i used a bread board and replaced the caps with resistors; the resistors were also the loads. i used resistors in series with the primary leads to limit current in the event of a short ckt. to the secondary.
during my experiments, once i was satisfied there were no shorts, i removed the resistors protecting the secondaries and the load resistors and installed caps.
i used a 24CT transformer and cut back the primary voltage to 60Vrms with a variac.
--DL
-
the CT diode bridge doubler is full wave at the top of the stack and at midpoint - no doubts on this one - just proofing.
OK, then this site is wrong:
there is an elegant trick which incidentally suits the old costly low-rated silicon rectifiers.
I think Figure 4B on this page has that elegant trick:
http://sound.westhost.com/valves/design2.html#s5 (http://sound.westhost.com/valves/design2.html#s5)
What are your readings for AC Ripple Voltage & VDC at the double and at the half voltage points?
-
Just to clarify things.
Which circuit in figure 4B is wrong, the 'Center-Tapped bridge' or the 'Voltage Doubler'?
There is no 'CT diode bridge doubler' in that picture.
-
jj - that site says:
The centre-tapped bridge is a very common arrangement, and no balancing resistors are needed. It provides a half voltage point, and ripple frequency is double the mains frequency on both full and half voltage points. The voltage doubler produces ripple at the mains frequency on the half voltage point.
That's consistent with other comments.
HTH
Chip
-
Which circuit in figure 4B is wrong, the 'Center-Tapped bridge' or the 'Voltage Doubler'?
neither are wrong.
There is no 'CT diode bridge doubler' in that picture.
there is - it's a center tapped bridge - fig. 4b left. sorry for the confusion - it's not a doubler.
--DL
-
No prob, I was just making sure that you guys were talking about the same circuit.
Since the screen is more sensitive to ripple, would it be beneficial to add extra RC-filter for it?
-
Which circuit in figure 4B is wrong, the 'Center-Tapped bridge' or the 'Voltage Doubler'?
neither are wrong.
There is no 'CT diode bridge doubler' in that picture.
there is - it's a center tapped bridge - fig. 4b left. sorry for the confusion - it's not a doubler.
--DL
Clearly, I'm totally confused. I suspect I'm not the only one. Can someone please accurately & succinctly state the "obvious" for me (and my brethren)? Thanks!
-
would it be beneficial to add extra RC-filter for it?
absolutely. LC would provide even better regulation, ripple reduction, but RC is easier on the wallet and chassis space.
-
VMS - if you look at DummyLoad's circuit from the other thread, you'll see that he uses a choke on both power rails - plates & screens. I don't know that a choke for the plates makes economic sense versus a simple C-R-C filter, but a choke probably is cost effective for the 1/2 voltage screen rail. Especially since you don't want to drop more voltage there through a resistor.
BTW what is that 6K resistor doing in that drawing? It would be dissipating 30 watts by my calculation! Does it simply represent a possible load on the circuit?
Chip
-
richard's ckt doesn't work. it shorts the secondaries. if you guys can come up with any other variations, i'd be happy to give it a go.
Compare the bridge orientations of Richard's drawing to the bridge orientations of your drawing using separate secondaries. You'll see the issue is the direction of the bridge. Arrange as is your drawing (but with 1 secondary) and it'll work.
But it's not really adding anything to use 2 bridges. You could use one bridge, and build out additional filtering as Richard's diagram already depicted.
-
Arrange as is your drawing (but with 1 secondary) and it'll work.
which drawing??? :dontknow:
--DL
-
Richard's first post to this thread.
But like I was saying... why dink with 2 bridges when one does the same thing, but easier, with less parts and fewer potential issues.
I think we're making this whole thing way too complex for what it deserves.
-
A follow up question:
Does the "center tap" need to be in the center of the PT secondary(ies)?
IOW could you use the full-wave bridge with center tap topology to create two power rails, one starting at 350 (a total of 250 volts for the secondaries) and another off the center tap starting at 245. One secondary would yield 175 volts AC and the second 125 volts AC.
Most VVR circuits only adjust the power amp voltages while leaving preamp (and sometimes PI) voltages alone. Use the higher voltage rail for the power amp and the lower voltage rail for the preamp. Would there be any advantage over having a fork in the power rail?
Just as an example, here's a torroidal transformer with a 240 volt secondary and a 220 volt tap on the same secondary. Not exactly the voltages I was thinking of but at least it's a PT diagram. Could you use the 220 volt tap as the "center tap"?
http://www.antekinc.com/pdf/AN-05T240.pdf (http://www.antekinc.com/pdf/AN-05T240.pdf)
Thinking about that Antek PT again - if the unbalanced "center tap" concept would work, you could have a 20 volt tap and a 240 volt tap. Might be useful if your amp has solid state circuitry, switching, etc.
Cheers,
Chip
-
In a way, this is how tube o'scope supplies are wired. I don't want to bust bandwidth and try to post a Tektronix o'scope manual here... But they have plenty hosted on BAMA (Boat Anchor Manual Archive) (http://bama.edebris.com/manuals/).
These supplies are regulated, and the regulated outputs are stacked one on top of another to provide various voltages (typically from -150v to 500v). The high voltage CRT supply is generally separate, but still has some of its supply components run from the stacked regulated output voltages. Page 4-14 of the Tektronix 545a manual shows such a supply. The output voltages are -150vdc, 0v, 100vdc, 225vdc, 325vdc (unregulated), 350vdc and 500vdc.
You could do what you described is you use bridge or full-wave rectifiers for each tap, then stack the resulting d.c. voltage components on each other. Let's say that's a 250vdc and 100vdc set of supplies. You might have the "ground side" of the 250vdc connected to the actual circuit ground, while the "ground side" of the 100vdc supply is connected to the positive side of the 250vdc supply, giving a total output voltage of 350vdc.
The key is you can never reference the "ground side" of the 100vdc supply to anything but 250v. Your options must always be either the 250v output or the 350v output. A single tube could be connected such that it feels as though it has a 100vdc supply; you simply have to keep every part that would be "at ground" for that tube and its associated circuits riding on top of the 250v supply.
-
> the higher the screen the higher the amplification
No.
Pentode is a current-limiter. G2 voltage sets the maximum plate current.
-
What happens if screens are left way to high? I built a Ampeg B15/Traynor YBA-3 using four 6550s. Plates and screens are 550V. I am using a choke with 1K 5 watt screen resistors. The amp will never be pushed hard. What would you do.
Thanks for any help
Rob
-
The 6550 thing is a recurring topic 'cause typical operation is 600V plates, 300V screens; and it's hard to get that 300V difference. Note that these are voltages at idle. Screens draw little current at idle. Ohm's Law tells us that a voltage drop = resistance X current. With little current flow there will be little voltage drop, no matter how big the resistance is. So, screen resistors "kick-in" under signal conditions, when the screens draw more current, but not much at idle.
Guitar amps tend to run 6550 screens at voltages higher than the spec sheets specify as typical (around 300) or even max (440). Hi-fi amps tend to use a UL arrangement for 6550 screens. The beauty of UL is that it solves the voltage problem at the source of the B+ supply, and doesn't require filtering (resistor, choke, cap) to the screen. However it is generally disliked for guitar, except bass.
I think if you run a search, a there are builders here have 550V on their 6550 screens.
-
The Major's screens are at 630v - which is probably why it eats the cheapo chine and many of the bloc tubes. KT88's or 6550.
"UL arrangement" "However it is generally disliked for guitar"
HEY, HEY, HEY!!!! Don't become a hater like 'Nit! :icon_biggrin:
Jim
-
Fender Champ 12 is a guitar amp that uses a voltage doubler with the half-voltage for the screen & preamps ,, that's how I got digging around a lot lately learning about what the screen in a pentode really does(6L6gc in this amp - single Ended! 496 volts plate, 242 V screen, cathode Bias ;) . After I reworked the"1980 high gain" into classic overdrive with an edge (when it's pedal switched to overdrive) and added more filtering & a choke, it has turned out to be the best lower volume amp I've ever had - and I was "put off" by the low SCREEN (& preamp) voltage - but it works extremly well (for lower volume/ 10 watts).
Here's the PS (C+ goes straight to the screen) and the complete stock schematic.
-
Is HBP still around? Haven't seen one of his posts for a wee while. (Seeing as how this 8-year old thread's been resurrected 'n'all)
-
He was last logged on on 4/23 but hasn't posted since last June. I didn't realize it had been that long. Don't know what's up?
-
This is a great thread and please excuse my probably ignorant question but assuming you had a power supply with two independent secondary windings, one for the plate and one for the screen, would it make sense to connect the screen supply to the cathode instead of 0V to have screen voltage as stable a possible? Assuming you want as much alternating DC-current in the cathode resistor as possible and no alternating screen voltage.
-
So that's how Ultra-Linear works.
silverfox.
-
that's how Ultra-Linear works.
You forgot, they make magic MOJO also :icon_biggrin:
-
that's how Ultra-Linear works.
You forgot, they make magic MOJO also :icon_biggrin:
YES!
Jim
-
aside;
My gigging step-Son is a fender amp kid :think1:, he wound up with a Marshall 18W clone and called last night all geeked about the sound n wanted to know why the difference. I explained about QUALITY IRON (and the rest) :icon_biggrin:
NOW he wants my old amps he test drove but wasn't quite happy with since they didn't sound fendery enough :icon_biggrin:
-
Very interesting but why bother trying to lower the screens B+ if the guitarists dislike it ?
The UL tap is not popular because the B+ for the sreens is too low, not before it is taken at a point near 40% of the opt winding. Guitarists like me don't care about replacing the tubes one or twice a year. They care about the tone.
If the tone would be a lot creamier and better and extraordinary with a much lower screen B+ amplifiers builders would drop UL trannies in their builds or modify the B+ supply accordingly.
In many of my builds, the screens voltages is higher than the plates. It doesn't affect the tone drastically to feed them with 5 or 10 volts over or under. Tubes life is more affected by bias than by supply, Imho.
Jack
-
... assuming you had a power supply with two independent secondary windings, one for the plate and one for the screen, would it make sense to connect the screen supply to the cathode instead of 0V to have screen voltage as stable a possible? ...
I think it would matter only in cases where you have a large cathode load/resistance, and can't use a bypass cap for some reason. An example would be the old McIntosh amps of the 50's because half of the tube's load is between cathode & ground, and they had to come up with some clever tricks to avoid the cathode-to-screen voltage varying by a hundred volts or more with signal.
"As stable as possible" brings up situations where you have to remind yourself to keep perspective. If you really mean "as possible" then you pay "all the money" and skip power supplies in favor of massive battery banks to provide the d.c. you want. :icon_biggrin:
Most likely, you have a push-pull output section so your opposing tubes can share a cathode resistor (and have closer to net-zero current change), or you've taken the advice from the tube data sheet to use individual cathode resistors, but heavily bypass each one.
So that's how Ultra-Linear works. ...
Very interesting but why bother trying to lower the screens B+ if the guitarists dislike it ?
The UL tap is not popular because the B+ for the sreens is too low ...
Huh?? How did UL enter the discussion here?
-
Huh?? How did UL enter the discussion here?
I thought the goal is to control the screen voltage off set. That's what UL taps do I think.
Now I'm wondering if a Mosfet in the 15ish watt variety would serve as a sort of follower and regulate the screen voltage at an off set? Sort of a tube regulator thing. Just an idea. Hi Watt does it with the PI; DL504, Kinda.
Nice to hear from you again.
silverfox.
-
UL is a very different thing to having a separate dropped HT supply for the screen grids.
UL acts on a tube’s dynamic conditions, effectively changing it into a different, triode-pentode hybrid tube type. A completely different set of characteristics and load lines apply to tubes used in UL.
Whereas the separate, lower screen grid supply thing just limits max plate current (and puts much less stress on the screen grids), the tube still operates in pentode mode.
-
Thanks for explaining all this.
-
UL is a very different thing to having a separate dropped HT supply for the screen grids.
UL acts on a tube’s dynamic conditions, effectively changing it into a different, triode-pentode hybrid tube type. A completely different set of characteristics and load lines apply to tubes used in UL.
Whereas the separate, lower screen grid supply thing just limits max plate current (and puts much less stress on the screen grids), the tube still operates in pentode mode.
Okay- So could a Mosfet be used as a sort of follower-regulator, for the screens instead of a resistor that introduced sag? Of course it would cost a little more. Could the Mosfet be employed to act like a UL tap?
silverfox.
-
I guess that might be theoretically possible but there’s massive voltage on a power tube plate, so a super tough device would be needed, along with some non trivial circuitry.
But mainly what would be the point? A tube amp needs an OT anyway, so if UL is desired then just get an OT with the taps.
-
> could a Mosfet
Can't boost the screen *above* the B+, as UL does.
-
Very interesting but why bother trying to lower the screens B+ if the guitarists dislike it ?
The UL tap is not popular because the B+ for the sreens is too low, not before it is taken at a point near 40% of the opt winding.
Jack
Colas,
I think you may be misapplying the 40% UL screen tap rating. Screens can be whatever you want in a UL application. Think of the 40% tap as the ratio between the triode/pentode operation, not a percentage of screen voltage reduction. Remember, if you connect screens to the center tap you have classic pentode operation. Screens connected to plate and we have triode operation. We are cheating connecting to the 40% tap (or other % of tube manufacturer suggested tap) and getting the best of both worlds resulting in less distortion and a more "linear" operation. There are actually quite a few "boutique" builders using UL designs these days. They tend to be very pedal friendly and also hi-gain friendly as they give you what you plug in to it and are not necessarily part of the distortion characteristics. This allows for the same audio result over the power/decibel capacity of the amp.
Jim UL fan!
-
Well, UL has a specific, technical meaning: the %age of tap being "optimum" for the tube type is UL. Yes, we can tap @ any %age we want; but this is no longer UL per se. This is because UL not only provides a certain screen voltage; but also does so in a way that provides a feedback effect which significantly reduces "distortion", while providing nearly full pentode power. If you deviate from that optimum point you are no longer ultralinear. For more info, and the optimum %age for various tube types, see: https://en.wikipedia.org/wiki/Ultra-linear (https://en.wikipedia.org/wiki/Ultra-linear)
-
Yeah, I wasn't real clear there.... I think(?) Colas was assuming the 40% tap meant a 40% reduction of screen voltage that he thinks is already low in a UL application. That is simply not the case (as I tried to allude to). For example, a bone stock Major has 630V on the plates and 629.9V on the screens at idle. Ok, maybe 628V but you get the picture. The tap and UL design has nothing to do with lower screen voltages. That is what I meant with "Screens can be whatever you want". Sorry, I was not real clear.
Jim still a UL Fan!
-
UL is a very different thing to having a separate dropped HT supply for the screen grids.
UL acts on a tube’s dynamic conditions, effectively changing it into a different, triode-pentode hybrid tube type. A completely different set of characteristics and load lines apply to tubes used in UL.
Whereas the separate, lower screen grid supply thing just limits max plate current (and puts much less stress on the screen grids), the tube still operates in pentode mode.
Okay- So could a Mosfet be used as a sort of follower-regulator, for the screens instead of a resistor that introduced sag? Of course it would cost a little more. Could the Mosfet be employed to act like a UL tap?
silverfox.
A number of ways to drop screen voltage have been discussed on this Forum and on the net, including mosfet circuits. I think the most satisfactory ways are UL, or a separate secondary screen winding. Other ways: diodes; zeners; attenuating series R; voltage dividers; mosfet or other regulator circuits. I think the problem is that (unless the plate supply is also regulated) these things tend to get the screens' voltage & current out of whack with the rest of the tube, under signal conditions.
-
...The beauty of UL is that it solves the voltage problem at the source of the B+ supply, and doesn't require filtering (resistor, choke, cap) to the screen...
Just to pick up on the above, I think UL doesn't really solve a voltage problem so much as a dissipation problem. By tying g2 to dynamically (partway) track the plate voltage, it avoids the high g2 dissipation condition of regular pentode operation, ie when at high signal level peaks, the plate swings down ~100V whilst g2 is being held up near its HT node.
I don't see that it's correct to think that in UL, g2 doesn't require filtering; rather that it's stuck with the degree of ripple at the OT CT node. It's still a grid and will have some amplification of the Vac it's presented with. It's a complex scenario and there may be some degree of cancellation due to the inherent local feedback mechanism. But to avoid high signal levels being modulated by HT ripple, it's best to supply the OT CT with a well smoothed HT, eg via a CLC Pi filter, as per the Sunn Model T https://el34world.com/charts/Schematics/files/Sunn/Sunn_model_t.pdf
-
Plate & screen supplies are generally spoken of in terms of supply voltage. Yes, reduced supply voltage causes a reduction in dissipation per Ohm's Law (including the Power Formula). This is a necessary implication, unless current draw were to somehow increase commensurately with reduced voltage, which it does not. So, this can be expressed in terms of any of the 3 components of Ohm's Law : watts (dissipation), voltage, or current across the impedance of the tube element.
I see no reason why a filter cap circuit could not be added to the screen UL supply. But this is not done, presumably because it isn't needed.
-
Plate & screen supplies are generally spoken of in terms of supply voltage. Yes, reduced supply voltage causes a reduction in dissipation per Ohm's Law (including the Power Formula). This is a necessary implication, unless current draw were to somehow increase commensurately with reduced voltage, which it does not. So, this can be expressed in terms of any of the 3 components of Ohm's Law : watts (dissipation), voltage, or current across the impedance of the tube element.
I see no reason why a filter cap circuit could not be added to the screen UL supply. But this is not done, presumably because it isn't needed.
Not sure why a reduced supply voltage is mentioned (presumably in regard of UL operation)? As UL is often used to apply a higher voltage on to g2 than regular pentode mode, the tube info supporting that, possibly due to the dynamic dissipation reduction with UL mentioned just.
https://tubedata.altanatubes.com.br/sheets/127/6/6550.pdf
As I see it, the use of UL taps precludes any additional HT filtering for the g2 supply; how could it be done :w2: If it was feasible surely the Model T etc would have that, rather than use 2 monster chokes to accommodate the HT current requirement of the whole amp?
I've read about very complex OTs that somehow have completely separate UL windings for g2, but those are surely a special case, outside of UL implementation as it is generally understood?
-
I've read about very complex OTs
If you've ever looked at the complex mathematics for solving complex, dynamic frequencies, normal sane folk will just plug in the ear buds n fire up a fat one :icon_biggrin:
-
Haha, I found EM theory modules the most difficult, worse even than laplace. Whatever, any competency I had in those topics, useful or not, is now long gone.
A redeeming feature of buying an OT is that someone else has dealt with the horrendous math involved in their design, and the tedium of their construction.
Leaving the customer free to do something more suited to them, whether that’s drilling a turretboard or sitting back as described :d2:
-
Huh?? How did UL enter the discussion here?
I thought the goal is to control the screen voltage off set. That's what UL taps do I think. ...
UL is a very different thing to having a separate dropped HT supply for the screen grids.
UL acts on a tube’s dynamic conditions, effectively changing it into a different, triode-pentode hybrid tube type. A completely different set of characteristics and load lines apply to tubes used in UL.
Whereas the separate, lower screen grid supply thing just limits max plate current (and puts much less stress on the screen grids), the tube still operates in pentode mode.
Okay- So could a Mosfet be used as a sort of follower-regulator, for the screens instead of a resistor that introduced sag? ... Could the Mosfet be employed to act like a UL tap? ...
Maybe the other tangents made it obvious UL doesn't do what you think it does.
At idle, the screen is at the same (or higher) voltage as the plate in distributed-loading (UL to you & me). But when signal is applied, it bounces all over the place just like the plate voltage. The screen voltage goes lower and higher, but because of the tapping along the winding the screen's voltage changes less than the plate's voltage.
If you're wanting a "stable screen voltage" than UL, or mimicking UL, isn't the way to go.
-
https://www.ampbooks.com/mobile/amp-technology/ultralinear/ (https://www.ampbooks.com/mobile/amp-technology/ultralinear/) (especially for Ritchie200)
https://www.diyaudio.com/forums/tubes-valves/189803-ultralinear-operation.html (https://www.diyaudio.com/forums/tubes-valves/189803-ultralinear-operation.html)
-
Very interesting topic, and funny to see it on top, I was just wondering if it was okay to set Vg2 higher than Va on purpose.
From what I read here, it seems to make no harm for small voltage differences (<10V), but what about 30 or 40V ? And what would be the practical limit ?
I need to double check, but I was looking at loadlines for a 6ak6 SE, and felt I would need to set Va around 140v, Vg2 at 180v to get the loadline lower than the knee. At the recommended point of operation, it goes right through it...
-
> At the recommended point of operation, it goes right through it...
That's usually near optimum. Why are you different?
FWIW, RCA had a super oscilloscope that ran 6L6 at Vg2=300V and Vp near 140V.
-
> At the recommended point of operation, it goes right through it...
That's usually near optimum. Why are you different?
Well, I thought the recommended way for guitar was noticeably lower than knee. It's the case on a champ for example, and I'm pretty sure I read this on Merlin Blencowe's book, or on his website. It was not quantified though.
Regarding the anode and g2 voltages relationship, given your example of the oscilloscope, I guess this is a non issue for the tube :-)
-
> I thought the recommended way for guitar was noticeably lower than knee.
Maybe. It may reduce grid-blocking. OTOH a higher impedance gets you near the same place. On the third hand you may not hit your power goal without exceeding some rating.
-
> I thought the recommended way for guitar was noticeably lower than knee.
Maybe. It may reduce grid-blocking.
From what I remember, it is to get more compression, thanks to the inrush current on G2 you get at very low anode voltages. It creates sag on Vg2, lowers the whole plate characteristics, and at lower Vg2, the unchanged loadline more or less goes through the knee...
I'll try to find a link to make sure I'm not totally off.
OTOH a higher impedance gets you near the same place.
Sure, but it also increases the max anode voltage swing. It's generally no big deal I suppose, but the 6ak6 I was looking at, it specifically has a reputation for not handling very well its maximum voltage ratings. That's why I was pondering this question.
-
Very interesting topic, and funny to see it on top, I was just wondering if it was okay to set Vg2 higher than Va on purpose.
From what I read here, it seems to make no harm for small voltage differences (<10V), but what about 30 or 40V ? And what would be the practical limit ? ...
"Practical limit" depends on the tube, and the voltage applied to the screen. The bottom graph on page 6 of this 6L6GC data sheet (https://frank.pocnet.net/sheets/093/6/6L6GC.pdf) shows you want to keep at least 25-150v on the plate, depending on screen voltage. To avoid high screen current:
- Plate should be at least 25v when 50v is on G2 (25v difference)
- Plate should be at least 50v when 150v is on G2 (100v difference)
- Plate should be at least 75v when 250v is on G2 (175v difference)
- Plate should be at least 150v when 350v is on G2 (200v difference)
But 6L6 has aligned grids and tries to encourage cathode current to fly past the screen & towards the plate.
... I was looking at loadlines for a 6ak6 SE, and felt I would need to set Va around 140v, Vg2 at 180v to get the loadline lower than the knee. At the recommended point of operation, it goes right through it...
I thought the recommended way for guitar was noticeably lower than knee.
From what I remember, it is to get more compression, thanks to the inrush current on G2 you get at very low anode voltages. It creates sag on Vg2, lowers the whole plate characteristics, and at lower Vg2, the unchanged loadline more or less goes through the knee...
If you're seeking maximum output power, going through the knee is the way to go; it's how RCA advises you to design a pentode/beam power stage right in the front-matter of their tube manuals.
From there, just be aware of the tradeoffs.
- If you use a lower-impedance load to rotate the loadline above the knee you reduce odd-harmonic distortion and lower output power, while steering clear of screen current rises.
- If you use a higher-impedance load to rotate the loadline below the knee, you increase odd-harmonic distortion (grid lines bunch together at both peak plate current and at minimum plate current) and lower output power. Screen current may rise in the area of peak plate current (which is minimum plate voltage), so screen resistors might be advisable in which case you also have the opportunity for causing compression (because screen resistor voltage-drop will dynamically lower screen voltage, which lowers possible plate current, which reduces output power).
The right balance is whatever you decide delivers the performance you want, while not destroying the tube.