Hoffman Amplifiers Tube Amplifier Forum
Amp Stuff => Tube Amp Building - Tweaks - Repairs => Topic started by: shooter on November 12, 2013, 04:55:49 pm
-
Here's my first winter build, most of the schematic is cobbled, re-works from others. Pick it apart before I get my trannies. I'm hoping it won't be a zero efficency amp! Inefficent is to be expected with SE. Once it works I'd like to put a clipping indicator on it. I'm building it with an eye to "mod" it to either a stereo, or my 1st PP build.
Thanks for the help getting it this far.
-
I don't understand why you put the 5 tubes in a small box around each other with the preamp tube in the middle of the 4 output tubes and their hi B+ when you have a very good sized chassis?
I think your going to have trouble. You might have made a ring of death for the preamp tube.
Brad :dontknow:
-
Not everything to change, but most of what I saw so far:
Have you picked transformers, or are those numbers notional?
250vac * 1.414 = 353vdc, minus ~1.2v for diode drops. Hammond also makes the 261M6 (http://www.hammondmfg.com/261.htm), which is 215v @ 269mA, giving 215v * 1.414 = 304v - 1.4v = ~303vdc. We'll start with this number for B+.
4x 12w tubes in parallel SE seems wasteful of tubes and sockets. I haven't shopped around, but Hoffman sells EL84's for $12/ea, while the 25w EL34 is $22/ea and the 30w 6L6 is $21/ea. Less than twice-cost for twice-power or more, and half the number of sockets. But let's say you have a compelling reason to use EL84's...
... I'm hoping it won't be a zero efficency amp! Inefficent is to be expected with SE. ...
Not zero, but low. You'll be idling the tubes at 48w total, because you might as well get as many watts as you can. Output would be 24w with perfect theoretical efficiency, you'll actually get somewhat less. Maybe 20w or so.
Idling at 100% dissipation gives 12w/303v = 39.6mA plate current per tube or 158mA for the whole output stage.
An approach to the ideal load is B+/Idle Current = 303v / 0.158A = 1917Ω
It looks like you might've peeped Hammond SE output transformers, and the closest match to 1900Ω and 158mA is the 1627SEA at 2.5kΩ and 160mA of unbalanced d.c. allowed. You might be able to go higher than that 160mA, as this OT is rated 30w 20Hz to 20 kHz and would likely tolerate a bit more current before any perceivable bass roll-off with guitar. In the end, maybe a toss-up with the 1640SEA at 1250Ω and 200mA unbalanced d.c.
My selected PT only has a 4A 6.3v winding, but the old Mullard data sheet says EL84 only draws 0.76A of heater current, so you still have 1A left over for your 6SN7 preamp tube.
You show cathode resistors at 100Ω per tube. Mullard's data sheet (http://www.mif.pg.gda.pl/homepages/frank/sheets/129/e/EL84.pdf) shows that with 250v B+, the EL84 flows 48mA with -7.3v of bias. 7.3v / 0.048A = 152Ω. You will want to experiment with cathode resistance starting with 220Ω, given you'll have a bit more plate/screen voltage and will likely need a hair more bias voltage to trim the idle current down. If you want the math behind my 220Ω guess, I'll show ya, but it's easier to start with that value and add parallel resistance to bring idle current up slowly. I figure these can be 1w resistors, but you'll probably feel better with 2w parts.
You should drop the value of the screen resistors, as the EL84 has a steep μG1-G2. Start with 100Ω up to maybe 270Ω; max screen current should be in the range of 10mA or less, so 1w resistors should be fine (but again, you can use 2w if you prefer).
You B+ bridge shows 12x MUR460's. These are 600v devices, and the PT I suggested has a peak output voltage of 215vac * 1.414 = 304v. In a bridge rectifier there are 2x diodes sharing the reverse voltage while switched off, and that voltage equals the peak voltage plus the voltage on the 1st filter cap. But, that total voltage is split between the two diodes, so each feels only the value of the peak winding voltage, 304vdc.
Summing up, you only need 4x MUR460's (one for each leg of the bridge) with a 215vac PT winding.
Your d.c. heaters are probably unnecessary, especially in a power amp situation. Dunno where you got 16v output for those, as 6.3vac * 1.414 = 8.9vdc, minus 2 diode drops will be 7.5vdc. This would still have a pretty hefty amount of ripple as 4700uF won't really sustain 4A of output. I'd suggest feeding the EL84's the raw 6.3vac with resistive balancing to ground. If you think you need d.c. heat for the 6SN7, source it from a separate filament transformer and a decent regulated supply (or at least a pi-filter to clean up the output a bit).
What's the switched 300kΩ from B+ to ground for?
Your 330kΩ from EL84 grids to ground is a shade too high. The data sheet gives 300kΩ as the maximum value for this resistor, per tube. So you need one of those for each or a single resistor of 75kΩ. Incidentally, that 75kΩ is also the load your input tube will be driving. The coupling cap should be sized for 75kΩ, so for -3dB at 70Hz you should have a 0.03μF cap or bigger.
BTW, the 27kΩ cathode resistor for the driver tube seems a typo. You also shouldn't bother with fixed-bias, or the unity-gain driver you asked about before, because you need that gain to "throw away" in your feedback loop around the power amp.
How did you arrive at the feedback circuit you have? You cannot bypass the cathode resistance of that first gain stage (at least, not the bottom resistor) because then you're monkeying the feedback.
The input stage needs a resistor from grid to ground.
-
I'll have to digest way more than the few minutes i'm stealin from my boss but to answer, explain some;
The trannies are ordered, Both are Edcor, pt is a 250v 200ma. The OT is 25W 1250ohm to 4ohm.
the PA section i stole from actually had 82r for the cathodes of each, i had 100r's so i went with that. The 1k screen was from a previous parallel version that ran 300r's at the cathode, acording to my musical son n his guitar playin friends "That amp is the shits". I played my digi music thru it and really liked the sound but the 12ax pre just drove the urine outta the parallel el84's, hence the 6sn7 but it's my first build with that tube n it don't look like the 12x family, it can deal with like 20ma plate, 200v dif from cath to heater n other oddites.
I'm shooting for just about maxin the el's drive without distorting the wavform out. I suspect i will be "adjusting" the pre til i get the right signal on the el's grid. The datasheet says AF Peak like 4.8 to 6v. I take that to mean 9.6 to 12vp-p signal on G1 of the el's to drive it to max ampfication.
thx for everything
-
The trannies are ordered, Both are Edcor, pt is a 250v 200ma. The OT is 25W 1250ohm to 4ohm.
250vac is a bit too much for class A EL84's. Single-ended operation, by definition, is also class A.
250vac rectifiers to ~352vdc; the Mullard data sheet I linked shows that the maximum G2 voltage is 300vdc (although I'm sure amps have run more).
But I gotcha... half the price for the OT.
the PA section i stole from actually had 82r for the cathodes of each, i had 100r's so i went with that. ...
I suggested I could show the math for deriving that cathode resistor. Let's take a simpler path...
Look for a minute at the Vox AC-30 schematic (http://www.el34world.com/charts/Schematics/files/Vox/ac301960.pdf). 4x EL84's share a single 50Ω cathode resistor. If each had its own resistor and you wanted to develop the same bias voltage (i.e., idle the tubes @ same current as with the shared resistor), you'd have 1/4 the current through the resistor so it would need to be 4x as large. Ohm's Law will confirm that.
That would put you at 200Ω, but 220Ω is the more commonly-available part now. If you get the ~350vdc you'e thinking about with your ordered PT, the cathode resistors would need to be even larger to trim the idle current down to compensate for the extra supply voltage. I figure you'll need just shy of 300Ω.
Again, I can show how that is derived. But know that from the 200Ω condition, you're raising plate which will require idle current to be reduced (which means more bias, so a bigger cathode resistor). Also, screen voltage will be increased with your higher B+ which raises plate current, which will then require the cathode resistor to be made bigger again.
Or you can try getting a pack of 100Ω and 50Ω resistors and making a series string of several-hundred ohms, then shorting out resistors one-by-one to land on your desired idle current.
... The 1k screen was from a previous parallel version that ran 300r's at the cathode, acording to my musical son n his guitar playin friends "That amp is the shits". ...
If 300Ω is the magic value, why change it?
The 1kΩ screen resistors will tend to cause screen voltage to sag when you push the amp towards maximum clean output. If you want sag (and slightly reduced power output) use the 1kΩ's; otherwise, you'll need smaller screen resistors to prevent power loss.
... I played my digi music thru it and really liked the sound but the 12ax pre just drove the urine outta the parallel el84's, ...
Are you sure you didn't just overdrive the input of the amp, or have too-hot an input signal?
... hence the 6sn7 ... I'm shooting for just about maxin the el's drive without distorting the wavform out. ...
Will this be a guitar amp or something to play the iPod through? Do you know how much sensitivity you want at the input jack?
... I suspect i will be "adjusting" the pre til i get the right signal on the el's grid. ...
I don't see a "preamp" but just a power amp and its driver. Previous question applies as to what is the intent for this amp.
If it's gonna be a guitar amp, are you gonna have volume or tone controls? If it's gonna be playing back your iPod, so you know what a median output is from your device?
Don't get me wrong, if you assembled what was shown in your schematic, you'd get a sound out of the speaker; no guarantees about how well it will perform...
-
> What's the switched 300kO from B+ to ground for?
Resistor probably a bleeder---- but why *switched*??
The whole amp draws 150mA-200mA. Bleeder sucks a whopping 1.1mA. That's like washing the car and a slow drip at the hose... not a problem. Let it drip the 1.1mA.
*AND* then you don't trust the switch to connect the bleeder when the amp is off. The resistor is HARD-wired, less to go wrong.
And why 10 Watts? You can't readily buy 300K 10W. And as-drawn it can only drain the caps.... hmmmm, could take *120 Seconds* to drain 400uFd down to 63%. So I guess it could get hot for a while. How hot?
The dissipation 300+V in 300K is under 1/2W left-on. 1/2W parts may be rated 300V max, and we're close, so I might stack two. Bleeder is important, so spend a few cents for 2W parts working 1/2W each. 160V at 1/2W is 51.2K each. You are already buying 50K 2W for plate-load, get more. Bleed-time is now 40 Seconds; I'd really like less. (It *will* be less, because the 6BQ5s will suck for a second after power goes off.)
DC heat to power-tubes is pointless.
DC heat to preamp tubes should be cleaner than a one-big-cap filter.
V1b seems useless, a left-over from a push-pull design. I can't see any virtue in it.
The NFB won't work as shown.
With cap connected right, there's still no NFB. Compute the amp gain. 20W in 4 ohms is 12V peak. You will probably run 8V-10V bias and peak grid swing. Output stage has gain about 1.5. V1b stage gain is unity. V1a gain is about 12. Overall gain 12*1*1.5= 18. NFB divider is 22K:100 or 221:1. Gain is *much* less than divider ratio. Negligible NFB.
Input sensitivity now is about 12Vp/18= 0.67Vrms.
Guitar-amp needs more like 0.020V-0.050V input sensitivity. This is not a guitar amp.
This is OK for "hi-fi", except a naked (no-NFB) pentode amp is not real hi-fi. NFB trades-off excess gain for fidelity. 6dB (2:1) NFB helps a little; sensitivity around 1.33V which is tolerable in hi-fi. Serious hi-fi runs 14dB-26dB (5:1-20:1) NFB (or more!). This puts us up near 6V input sensitivity. I have done that, but the preamp must work HARD.
I think V2b should be wired as a gain-stage. Even then, and with no NFB, it's not a hot guitar amp. As a hi-fi amp, the added ~~13:1 of gain can all be turned to NFB giving excellent damping and low THD numbers. The NFB divider could be 1.2K:100r.
-
> 330k from EL84 grids to ground is a shade too high.
I'll wink at that.
He does say "6BQ5". The GE sheet under Cathode Bias shows 1Meg max (per grid). So 250K.
However few tubes get in trouble _at_ the datasheet Rg rating. Most leak a lot less. Unless he gets gassy junk, at-most one tube needs 1Meg, the others could be stable with 2Meg or 3Meg. So 330K for four is not likely to cause trouble.
OTOH he's driving from 27K or 50K. And 6BQ5 does not need huge drive. 250K (220K-270K) is not a problem. And a wee bit more leeway if he does get junk tubes.
-
The amp will be my experimental amp, mostly for music. No knobs for now cuz all my upstream devices have vol, tone, eq. but none of them have “eyes :dontknow:” I’ll be scoping this one a lot. The previous builds I gave up scoping cuz my kid and friends would say, “awesome” and my scope would say “excrement!” I’m building an add-on tag board with 150r’s to add to my 100’s and 1k’s to parallel the screens if the currents get to hot. The switched 300K is a power rail bleeder, think I got the value off a guitar amp schematic and I have a few 150k 10w’rs just layin around. There was a comment about chassis layout, it was done as a visual artist. I’m shooting for like 10v out of the 6sn7 with shielded inputs and outputs, do the tubes cross-talk from mechanical proximity? Or create emf fields like trannies? This build has nothing to do with plain blue collar logic, it’s winter and I need something to do! If you wanna see impractical, check out this amp, think it said 30,000 cost for 30 watt stereo! BUT she is just amazing!!!!
http://www.theaudioeagle.com/thomas_mayer_tube_amp_designs.html (http://www.theaudioeagle.com/thomas_mayer_tube_amp_designs.html)
anyway, I have another 4 weeks or so til my iron shows up in a semi! So build up some of the suggestions then I can swap til I like.
Thanks for everything!
-
I have one thing about tubes I keep missing, and this build has 3 different examples,
that is the grid, g1, the gas pedal. On the input, most guitar amps have like a 100k to grid with a 1 meg to ground. My input came from 2 dif audio builds, one had a 56k, the other a 100k. The 2nd half of the 6sn7 holds the grid at a fixed voltage, make B3 200, the r ratio is like 4/1 so the grid sees 50v (ballparkin the no#’s). The 6bq5’s also fix the grid but only above ground(330k) (i see the 10k as like a fixed volume input). My trippin up part is how do each of these apparently different configurations keep the grid neg in respect to the cathode? I think I’m in the neg 7 or 8 range on the el’s datasheet but I have no clue how to translate that info and make it so with resistors!
-
My trippin up part is how do each of these apparently different configurations keep the grid neg in respect to the cathode?
The answer is the same for all three examples... The cathode has a positive voltage that is greater than any voltage on the grid. So, the cathode is more positive than the grid. Said another way, the grid is less positive than the cathode. And that's the same as saying the grid is negative with respect to the cathode.
-
And that's where i'm missing the boat or something, the el's grid is setting 330k above ground, the cathode is only 100 above ground, how does that make the grid neg with respect to cathode?
-
... the el's grid is setting 330k above ground, the cathode is only 100 above ground, how does that make the grid neg with respect to cathode?
Ohm's Law. Voltage = Current * Resistance
EL84 cathode has couple-hundred ohms to ground, but direct current is 10's of mA.
EL84 grid has couple-hundred-thousand ohms to ground, but direct current is in the pA (pico-ampere) range.
Voltage across cathode resistor is 10v or less, but voltage across grid resistor is substantially-zero.
-
And that's where i'm missing the boat or something, the el's grid is setting 330k above ground, the cathode is only 100 above ground, how does that make the grid neg with respect to cathode?
Use your voltmeter. Put the black lead on the cathode, and put the red lead on the grid. What do you see?
-
N the dim bulb grows brighter! N on one build i did make that measurment and it was as the datasheet showed - sorta. In my out of phase mind i was seeing the grid n cathode equal 'cept for the resistance. thanks
-
I call it QB-22. The good news first, She sounds sweet! The amp is my 2nd audio-build, very tight bass, good roll-off on the treble, just before it becomes “shrill”, and plenty of volume. I have the equalizer flat with a 3db bump on bass. I have one cranky tube, but it’s workin the kinks out. 1V in gets you 340v at the output tranny and 18v at the speaker. I’ve attached the dc data, initial, and after 10 hrs. The pwr rail resistor is under watted, so she’s nearing failure, hence most of the changes over time.
Todo;
Get a bigger pwr rail resistor, chassis mount 50W, also see if there’s a way to get more gain out of the driver V1a. Add a GNB for experimenting, switch for changing from UL to straight pentode, finalize the schematic. All in all was a fun build, maybe not the standards of the pure audio guys, but hey, I have zero hearing above 12Khz so why get to picky! I learned a lot about the design end…THANKS!