|23rd September 2005, 04:04 AM||#1|
Join Date: Feb 2001
Location: Columbia, SC
My big mistake was admitting that I had ever done tube work.
My somewhat lesser mistake was doing something foolish like posting in the Monster thread.
The straw that appears to have broken the camel's back was when Coulomb requested an "X" tube preamp...and I went and said,"Sure."
Oh dear, Gussie...I dun went and let myself in for it...
Okay, the story began this past January in one of the Pass Monster threads. (I think there's more than one.) Anthony (aka Coulomb) said something about an X preamp--I think he specified tubes--and I said I had done a bit of fiddling along those lines. Well, wouldn't 'cha know, about three or four posts later, Nelson Pass (aka Nelson Pass) up and says,"Hey, bub, you're always talking about how tubes do voltage like nobody's business...well, this Monster thing is going to take a bit of voltage, so why don't you jump in here and do a front end?"
So I, (aka GRollins) undertook to do so.
There's just one little, tiny, itsy-bitsy, teeny-weeny fly in the ointment. Nelson wanted 160Vp-p and 50mAp-p at the output of this front end. Now, anybody who's ever messed with tubes will tell you that 160V is child's play. But they'll also look a wee bit troubled when you get around to mentioning current on the order of +-25mA. I mean, really, that's power amp territory, you know?
There's something you've got to realize. Tubes and current...they ain't such good friends.
I decided that I should approach this somewhat in the spirit of the way Nelson does things. That means a single gain stage, if possible. Did I mention that he specified cascoded and bootstrapped? No? Well, it must have slipped my mind.
I don't much mind cascodes. Not so much as I used to, anyway. But bootstrap circuits give me hives. Why? It's positive feedback. That makes me nervous. Us poor, dumb, Southern boys get 'bout near as skittish about that stuff as when some brassy Yankee filly comes along and starts suggesting that we do things the way they do it in foreign places, like France. Private things, if ya catch my drift.
Well, if you say so, Boss.
The first thing to decide was what tubes to use. I set a couple of constraints. One--the tubes must be in current production. Two--the tubes must be likely to remain in production for the forseeable future. I also set myself the goal of using triodes. Why? Cuz I like 'em.
Okay, the obvious front end candidates were the 12AX7/7025, 6DJ8/6922, and 6SN7. To make a long story short, the 12AX7 can do voltage, but not current. The 6DJ8 can do current, but not voltage. The 6SN7 can do both, albeit with less gain. Such is life.
I was dead set on using the 6AS7/6080 power triode on the output, but a little calculation showed me that it was going to be a no-go proposition. The voltage and plate dissipation were both going to hit me at about the same time. I decided not to mess around. I pulled out a quartet of 6550s. Huh! That'll show 'em!
6550s? Hey, Grey, that's not a triode!
Sure it is, goofus. It's just got this pesky extra pin (aka Grid 2) that you've got to attach to the plate with a 100 Ohm resistor. I thought you knew that.
Quoth you,"Oh, okay."
The rest, as they say, is history.
Mind you, I could have laid this thing to rest in about two weeks if the real world had left me alone, but that never seems to happen. As it worked out, it took nearly nine months to gestate this critter and put it out for common consumption. That's all right. The output stage/chassis guys got all hot-'n-bothered about CLC power supplies and such. It's not like they were in such a hurry for the front end.
Okay, on to the circuit.
The front end is a cascoded differential. It's biased by a current source. There's a pot, V5, that is to be used to fine-tune the bias. It's the only adjustment in the whole circuit and it's really very simple: Turn V5 until the plates of V4 and V6 are at 200Vdc. There...you're done.
That current source is tied to the negative rail of the filament supply for the front end tubes. That filament supply is to be 6.3Vdc filtered and preferably regulated. Tie the filament positive to ground. This leaves the filament negative at -6.3Vdc. Presto! We get double duty out of the filament supply. It heats both 6SN7s and anchors the front end CCS. Neat, eh? (Actually, I'm just lazy. This is the electronic equivalent of a computer programmer saying,"That's not a bug, that's a feature!")
The output tubes, the 6550s, need a separate filament supply. Why? Well, tubes don't like to have too much of a voltage difference between their cathodes and their filaments. Bad things happen if you exceed the ratings. Baaad things. Don't do it. Instead, you create a separate 6.3Vdc supply and tie it to the midpoint of a voltage divider between the rail and ground. Doesn't matter if you make it +6.3V, -6.3V, or the midpoint of the 6.3V (should you happen to use a filament transformer with a center tap). A couple of volts either way isn't going to be a problem.
There's a funny thing about this output. Nelson asked for +-50mA. I went so far as to call the man up and press him on the matter. The conversation went something like this:
I said,"Okay, so how much current did you want at the output?"
Said he,"Oh, whatever I put online."
"I think that was +-25mA."
"Then that's what you use."
"Um, well yeah, but didn't you want a little extra? The 6550 is a beefy tube. It can take it."
"Just the +-25mA."
Okay, so I'm dumb. It took me a bit to figure out where he was coming from. He was expecting a little distortion when the thing was going full blast. (To tell the truth, so am I. At full volume, I expect this thing will roust dead bodies from the cemetary a half-mile down the road. [Roll over Beethoven...]) But, you see, I was trying to get him to advance some sort of rule-of-thumb as to what percentage of the quiescent current he expected to use, so that I could turn that around and use it to provide an appropriate Iq. Eventually, it dawned on me--the man intended to use all the standing current. When he said he wanted +-25mA, he expected that the output stage would be biased at 50mA, and that it would run all the way from pinch-off to clipping.
Being me, and some of you may have figured out by now that I like elbow room, I went ahead and programmed the output stage for about 60mA of bias. So we've got a bit of leeway. The 6550 can take it. For that matter it can take more, but keep an eye on the plate dissipation.
For that matter, you can put an output transformer on the blasted thing and listen to it as an amplifier in its own right.
No, I'm not. Use efficient speakers and the right ratio in the transformer and you'll have music.
Anyway, I see that I haven't mentioned that the output portion is a cathode follower (think Source/Emitter follower) biased by a current source. It is. In fact, I purposely set this thing up so that hard core solid state folks can look at it and pretty much figure out what everything does, even though it's made out of vacuum tubes.
Can you do this other ways? Oh,yes. I might even take a whack at another front end later on if I have time. Another tube front end? A solid state one? Who knows? It depends on what mood I'm in and how much time I can scare up. (Hey, Halloween is coming, smile.)
As it stands now, the circuit has somewhat over 26dB of gain. It's -3dB at a little over 100kHz. I was going to measure distortion, but I've got some residual fuzz due to the fact that I don't have enough capacitance on the power supplies. That artificially inflated the reading. That said, I ran the output up to 20Vrms (in order to swamp the noise) and got a THD measurement of around .35%. That's not too bad, considering the modest amount of feedback used. (Gimme a break, guys, that's 50W into an 8 Ohm load!) If I can remember where I put the box with my spare high voltage capacitors, I'll pull 'em out and augment the power supply. Then I'll do a proper distortion measurement.
It'll pass a pretty decent 10kHz square wave. My prototype shows a slight overshoot, well damped. Not even a full cycle's worth. If you're bothered, put smallish silver mica caps in parallel with R6 and R24. A few pF should do the trick.
I talked it over with Nelson, and he feels that the Zeners at the Gates of the output MOSFETs will be sufficient to handle the turn-on bump without having to go to relays. Good, that's one less thing I have to monkey with.
Now, if you fellas (and gals) don't mind, I'm going to try to get some shut-eye. I'm on short rations and I want to be bright-eyed and bushy-tailed tomorrow when I pounce on that phono stage I've been thinking about.
(Note: read and approved by Saint Cindi, possessor of a tin ear and lots of patience!)
P.S.: My wife...I think I'll keep her...
|23rd September 2005, 04:54 AM||#2|
Join Date: Mar 2003
Location: St. Catherine, Jamaica
Dude thats a crazy circuit for a driver.
For one channel thats enough for a 100w tube amp.
So I need 8 6550 tubes ($$$$$$$$$$$$$$) for a stereo driver, thats crazy
With this driver's power supply, the power stage power supply, heat sinks and chassis this Monster will easily weigh 200lbs.
Is someone designing a SMPS for this project????????
<a href="http://www.danasoft.com"><img src="http://www.danasoft.com/sig/lawbadman.jpg" border="0"></a><div style="font-family:arial,sans-serif;font-size:11px;"><p>Sign by Danasoft - <a href="http://www.danasoft.com">For Backgrounds and Layouts</a></p></div>
|23rd September 2005, 05:24 AM||#4|
Join Date: Feb 2001
Location: Columbia, SC
Crazy, the man says...
As it happens, I agree.
It's my understanding that they're attempting to build a 1000W amplifier. Something really huge, anyway. It's been a long time since I subjected myself to that sort of punishment.
The problem that you run into is the summed Gate capacitance of all the gazillions of MOSFETs in the output stage. That capacitance has to be loaded and unloaded very quickly indeed in order to get reasonable performance from the circuit. That leads to the output current that Nelson requested.
The voltage swing is, literally, only half the actual output. The amp is bridged, so it doubles the voltage swing, once for each half of the amp.
I don't think I said so above, but the measurements were all single-ended, with the undriven input grounded. I didn't bother setting up a phase splitter to drive the thing with a balanced signal.
|23rd September 2005, 06:04 AM||#5|
Join Date: Jul 2003
Location: San Diego, CA
SmarmyDog is very pleased indeed! I was worried that my lack of progress reports had eroded enthusiasm for this project, but I’m ecstatic so see it’s moving along.
The Pass Monster has been discussed in a few different threads so here are a few (dated) links to pictures and PDF’s to orient the uninitiated:
Power Supply Schematic.
Power Supply redrawn based on parts-on-hand.
Output Section schematic.
Concept sketch (front view) showing three-chassis design.
SolidWorks rendering showing general location of capacitors and power trafos.
Photo of WIP sometime in June.
Another June photo.
One of my 4 year old twin sons has hernia surgery quite early tomorrow morning so I’m off to catch some sleep, but I’ll post a status update tomorrow…
|23rd September 2005, 11:54 AM||#6|
Join Date: Feb 2001
Location: Columbia, SC
Sorry to hear about your son. Hope all goes well with his surgery.
My desire for sleep has not worked out, so I'm running on about three hour's worth at the moment. With luck, you'll do better.
The list of schematics and such should prove useful.
|24th September 2005, 12:37 AM||#8|
Join Date: Feb 2001
Location: Columbia, SC
No doubt. I keep meaning to put together a small PCB whose only function is to phase split. So far, I haven't gotten around to it. No rational excuse except lack of time. As I get further towards my goal of a balanced system, I'll need it more and more often (my signal generators are both single-ended), so I'd better get moving.
|24th September 2005, 03:41 AM||#9|
The one and only
I recommend that you save up your shekels and buy at least one
of Jensen's dual primary/secondary line level transformers. You
can't beat them for passive conversion and/or isolation when
you want to look at something differentially or drive something
differentially. My AP System 1 uses them, and they cost maybe
$40 - well worth it.
|Thread Tools||Search this Thread|
|Thread||Thread Starter||Forum||Replies||Last Post|
|"Monster" Cable vs Monster Vintage||MonsterVintage||Everything Else||0||19th October 2004 12:35 AM|
|New To Site?||Need Help?|