PSU observations and questions

[daz]

i'm trying to understand the way the node R values are selected and trying to figure out what i'm seeing in the 2205 and 4210 marshall schematics. I keep thinking i have to be spacing when i look at these amps because i don't see any PSU nodes for each preamp tube. in fact, it looks as tho after the PI the PSU has not nodes separating the 3 preamp tubes. (2 of the 5 are reverb and PI) What am i missing here? I was reprimanded over at ax84 once when i showed a schematic with 2 tubes on one node. yet unless i am spacing i'm seeing worse than that in these marshalls. The only thing i can think of is they are diode clipping amps and maybe they don't need much filtering because they are using SS overdrive instead of tube? Someone enlighten me please.

Second, can anyone explain to me a way to determine what resistor values to use at each node without electronic mathematics. Because someone already told me and i have to tell you math and the simplest equasions send my brain spinning. It's my absolute Achilles's heel as mental aptitude goes. i don't want perfect, just wanna see if i'm in the ballpark. If not, would anyone be able to if i post a schematic or post voltages etc?

[Enzo]

Because a B+ node separate for every triode is simply overkill and not necessary. This is not NASA or laboratory equipment we are building here.

Rather than listen to the online people tell you that what you see in hundreds of successful commercial designs is wrong, perhaps you might look at a bunch of commercial designs for an idea of what is necessary. They won't put more stuff in an amp than is needed. And unless we find a particular amp model just doesn't work, we can assume for most purposes that their power supplies were designed well enough.

And yes, I see the irony in my being a guy on the internet telling you to ignore the guys on the internet.

The first couple filter caps in a power supply are just that - they filter out the ripple to make smooth DC for the plates and screens of the output. The nodes for the preamp also have filter caps, but their purpose is more about decoupling than filtering. Decoupling is I suppose the filtering out of remnants of signal though.

The point of decoupling is to keep signal out of the B+ so it won't get into other stages - we don't want the stages interacting. So we don't want the signal riding the B+. But if the related preamp stages are not in a relation that fosters interaction and/or the decoupling is stiff enough, then multiple nodes is not necesssary. There is nothing cosmic here.

SOme amps may have several triodes off one node, but if they are channel switching amps, then only some of those triodes are ever in use at the same time, depending on channel selected, as an example.

As to the values, they are not critical. You can start with a general idea of the current through each triode - a ma or so - and then just decide what sort of B+ voltage you want for that stage. The resistors between nodes are there as part of the filter, but they also are selected to drop specific amounts of voltage. If the PI runs on 400v and you want the preamp running on 300v, and there are 4 triodes in that preamp, then I have to drop 100v with about 5ma in a resistor. In this example then, I'd need 20,000 ohms. R = V/I = 100/.005 = 20,000.

And math/smath, that is simple arithmetic. If you learn anything in electronics, learn that. it is Ohm's Law, one of the most fundamental concepts in electronics. Can you look at the miles driven on your car and the amount of gas and come up with the miles per gallon? That is just as complicated. If a 6-pack costs $4.50, can you figure out what each can costs?

It is not like we are calculating some frequency response at these B+ nodes. If you change the resistor, it will just raise or lower the B+ at the following node some amount.

[daz]

Well, when a bunch of people gang up on you and tell you to "listen to him, he knows more than....", then it's hard to assume otherwise till someone like you says different. I trust you more than most, but you weren't there when that little debate took place. In any case, i guess theres no way i'm going to figure out what would be optimum because i don't even know how much voltage i want at each stage. I was just just curious anyways. Not like i need to mess with it, but i just got to thinking how it's about the only part of the amp left i never really understood to a reasonable degree and the values i used were just more or less arbitrary .

[Steve Conner]

Hi guys,

I agree that 2 tubes on one node is no problem, and I'll try my best to explain why now.

The decoupling Rs and Cs serve three purposes. The obvious one, that we all know about, is to filter out hum. The raw B+ at the plate and screen supply nodes of the power tubes, while fine for the power tubes, just isn't good enough for preamp tubes. You want as many cascaded RC stages as possible between the screen supply node and the first preamp tube, since it's the most sensitive. Most classic amps have two stages.

The next most obvious is to drop the voltage to something more suitable for preamp tubes. The CBS silverface Fenders dropped hardly any voltage and ran their preamp tubes off nearly 400V (you should show those schems to the AX84 guys!) and they don't get a very good reputation for tone. People seem to think that preamp tubes sound better at about 250V, and 250V was one of the recommended voltages in the 12AX7 datasheet, so the designer would have chosen his resistors to shoot for that. Maybe he'd have liked 350, but the resulting resistor made for a short enough RC time constant that it didn't filter enough hum.

To calculate the values, just reckon 1.5mA per preamp tube and use Ohm's law to work out the volt drop. The answer is usually 10k or 27k, so I just buy a bunch of 10k and 27k 2 watt resistors and try either

The most subtle is decoupling. This means stopping signals from feeding through the B+ rail between stages. Again, two triodes on one node is no problem, provided that they're parallel (like the two inputs in a 18W Marshall) or successive stages. With parallel stages there can only be crosstalk, and like Enzo says, we're building a guitar amp, not a radio telescope, so who cares about a bit of crosstalk. With successive stages, any feedback through the B+ must be negative, so it can't cause instability. To get instability you need two non-successive stages (ie, with another inverting amp somewhere else in between them) or three successive stages.

I only ever had trouble like this in one homebrew that had four successive triode stages all running off the same node

[cbarrow7625]

Daz,

Enzo & Steve have given you some good, real world advice and a couple rules-of-thumb to start with that are right on. It's not that you can't put multiple valves on the same node, it's just that you have to be careful about how many you put on any one node if they are part of the same signal chain. Signals bleeding back into the B+ that are in phase with each other can cause oscillations (very tricky to find if you don't know where to start looking). Two tubes in series in a preamp with opposite polarity on the plates (the tubes are inverting after all) can be safely run off of one node in pretty much all circumstances,even in a hi gain amp. If it is a low gain circuit, you can even get away with 3 preamp sections on the same node but I would not try that with a hi gain amp - you'll be asking for oscillations a lot of the time.

I try to stick to having no more than 2 tubes in series off of the same node, but make sure the plates are opposite in polarity. If you put stage 1 & 3 of a hi gain amp on the same node, you will have in-phase signal bleed back into the B+ (since the output off of each of those plates is basically in the same polarity) & it will likely cause you problems.

Voltages on the preamp tubes really just determine where you run out of headroom and distortion begins to set in. Typically, lower voltages = earlier onset of distortion, higher voltages = more clean headroom with later onset of distortion. Lower voltages tend to make the tone feel "softer" and hihger voltages make the tone feel more "stiff".

Fenders with lower voltages typically have a pretty good feel. The CBS-era, with their high voltages, are pretty awful to play on. In contrast, a hi-gain amp with low voltages can sometimes feel too "mushy" with no distinct pick attack present. Higher plate voltages for this type of amp can actually sound & work better than fender-type voltages.

I built an amp that contains both a Fender-type preamp & an SLO100/DR type hi-gain preamp. I got really confused at first because I couldn't get either one to sound the way I wanted it to. I finally realized that I had to run them off of completely different voltages. I lowered the voltages to the Fender circuit & raised the voltages to the SLO circuit & now everything is right where I like it. To get the voltages dialed in I actually put screwdriver-adjustable pots (around resistors) into the B+ lines so that I could make quick changes to the plate voltages to compare tones - this helped a lot. I wouldn't recommend leaving the pots in after you get it where you want though.

I think the best text I ever read on preamp voltages& nodes & how they interact was in Kevin O'Connors "Ultimate Tone" Series (maybe in the first book? - I'll have to go look). I highly recommend them for getting your head wrapped around these things. Much more informative than most of us idiots (myself included) telling you their opinions on a message board. You'll get good facts there & be allowed to make up your own mind. If that doesn't work, the rest of us idiots will still be here to help.

[daz]

Guys, that was excellent ! Thanks for explaining things in a way that i could understand without the typical bashing i get elsewhere for not being able to do the math. I have a serious problem with electronics math. I really do. And often people get rather pi$$ed when i tell them that and in so many words tell me to f off if i can't figure it out. I'm not trying to become another Enzo, i just wanna make sure my amp is sound and correct before i close her up for good. (yeah, like i'll be able to leave it in the cabinet)

Oh, and one more....the amp has a CF fed tone stack, and just as i see in other amps like that, there is no plate R at that stage. So that side of the triode is at 300+ volts. Why is this? Just curious because as was said, high voltages tend to sound stiffer and there seems to be no option to lower it because i think it needs to be like this for the TS. Any thoughts?

[daz]

I gotta tell you, youse guys are good ! I really just wanted to make sure i had the PSU right, but i tried what Steve mentioned about the 27k and 10k R's. man, i really didn't think the amp could get much better but that really tuned it in ! Sweet sweet sweet ! I happened to have a few 27k's and i used them on the PI and last preamp tube nodes, then threw the 10k on the first node. the feel is just squishy as all get out and the tone is even richer. I had no idea ! Thanks all.

[martin manning]

What you are hearing is the effect that cbarrow talks about- lowering the plate voltage moves the operating point down and left on the tube characteristics, reducing headroom and bringing distortion on at lower signal levels.

In the attached plot you can see this. I assumed a 100K plate resistor and a 1500 Ohm cathode resistor for 250V and 350V supply voltages, then used Ohm's law four times to plot two anode load lines and one cathode load line:

I=V/R, so 350/100K = 3.5mA, and 250/100K = 2.5mA for the anode load line intersections with the y-axis (where all the supply voltage is dropped across the plate resistor and the current is maximum). If the plate current is zero, then there is no voltage drop and the plate voltage is equal to the supply voltage, giving the intersections with the x-axis.

Assuming a 1V and a 2V bias via a 1.5K cathode resistor and using I=V/R again I get 1/1500 = 0.667mA, and 2/1500 = 1.33mA for the end points of the cathode load line, which are plotted by following the 1V and 2V Ec lines to the calculated plate currents.

The operating points are circled, and you can clearly see that there is less grid voltage swing along the anode load line between saturation (Ec=0) and cut-off with the lower plate voltage.

MPM

[daz]

That makes sense looking at it from my novice perspective. This is apparently why i had to really make the stages hot to get enough drive, then they would get harsh. If i removed enough gain to lose any oscillation and/or harshness it would have very little gain. So this way i don't have to slam the signal so hard to get the goods and thereby it's a sweeter distortion. Even the issues the hammond OT i added seemed to have cured are now 110% gone. I didn't even realize there was any of that left, but upon doing this to the node resistors it became evident there was a good bit more tone and feel to be had. It's frighteningly close to my idea of perfection. And to think before i started this building thing i was about ready to pull the trigger on my 4th JCM. Never again....I make em better than marshall !!!

[Steve Conner]

man, i really didn't think the amp could get much better but that really tuned it in ! Sweet sweet sweet !

Cool If you've hit a sweet spot, it might be a good idea to measure your preamp supply node voltages and note them down.

[daz]

Well, i don't know that i've hit THE sweet spot, but it's definately sweet. But i'm not naive enough to this i happened upon the perfect values, so i will likely try others.

[daz]

Whoa....i never check the voltages till now and they are scary low ! 108 on one of the side of V1. How can it sound good like this? It cleans up just fine. Is this bizzare or what? I suppose i'll try some small R's, but i'm not sure whether to go by my ears or what is theoretically correct. As far as i know it shouldn't sound good like this, but i do know tone and it sounds far better than it did with a couple hundred on the lowest triode.

[tubeswell]

Well my understanding (much enhanced over recent months) is that the amount of distortion is not related to the plate voltage alone. It also depends on the tube bias. If the tube is biased so that the grid voltage is somewhere near the middle of the load line (say for a 12AX7 -1.5V), the signal sine wave shape will be more similar for each opposite side of the voltage swing than if the bias point was more negative (say for a 12AX7 - 2.5V), or more positive (say for a 12AX7 -1V)). If the bias point goes too positive (nearer to 0V), then 'grid current limiting' will cause the postive voltage swing on the grid to clip the signal, and if the bias point is too negative (say -3V), then the negative voltage swing of the grid signal will cause the plate to clip. Similarly if the voltage signal you were pumping through the triode was too high, the signal would clip on both sides.

So you can have a lower voltage and a clean signal depending on the bias point. The sweet spot occurrs where the bias point is such that on the positive voltage swing side of the grid signal the triode amplifies the signal slightly more than on the negative side of the voltage swing, i.e. a slightly uneven voltage swing comes off the plate, compared to what went in on the grid. This produces the harmonic distortion that gives the valve its characeristic sound.

[daz]

Do you need a scope to measure that?

[tubeswell]

if you are going to measure your existing build then yes you would need a scope to see the differences in the sine wave shape. But you could hear the difference with a signal generator or just playing your amp, esp if you setup the cathode resistor such that it was adjustable to allow you to shift the bias while you were experimenting. (Altho' adjusting pre-amp valve bias by twiddling with a variable cathode resistor is noisy - so you wouldn't necessarily want to make a variable cathode resistor a permanent feature - unless you were going to set it and leave it kind of thing)

The way to calculate it theoretically is using a static anode characteristics graph for the type of valve you are using, and putting the load-line on it that has been calculated in accordance with the voltage of the plate and the plate resistor load. - Explained in this document:

http://www.freewebs.com/valvewizard1...Gain_Stage.pdf