Audio Asylum Thread Printer
Get a view of an entire thread on one page
|For Sale Ads|
In Reply to: Resistor as parafeed plate load? posted by str8aro on October 28, 2001 at 19:38:06:
Sure you can use a resistor. Try it. It will have a lower output impedance to the primary than choke loading will provide. It's not efficient for power output stages, but for interstage it's a very viable alternative.
I have a nice little test bed amp using 71As a 400vB+ and 10k plate resistor cap coupled to a rat shack line tranny gets .7 watts and nice bandwidth. Without a cathode resistor bypass cap output goes down to .5 watts but looks nicer. I've yet to listen to this rig it was intended to drive input transformer phase splitters.
I think it's useful for low power apps or where a high B+ transformer is all you have the resistor might get hot and needs dissipation attention.
Actually this is for a power output stage, probably a 2A3. When you say it's not efficient, are you talking about all the heat the resistor would dissipate? I was planning on using a 3K resistor with a 2A3 at 60mA, so that would dissipate about 11 watts.
> When you say it's not efficient, are you talking about all the heat the resistor would dissipate? I was planning on using a 3K resistor with a 2A3 at 60mA, so that would dissipate about 11 watts.
The resistor dissipates a lot. Also, the load on the 2A3 you have here will be the parallel combination of the output transformer primary and the 3K resistor, down around 1.5K. That's also going to make it more nonlinear and deliver less output power than using a choke. It can work, though, and make music.
i don't want to open a whole can of worms here on loadlines and power out, but isn't Rl=2Rp considered by many as the case where you get the maximum undistorted power output from triodes???
those aren't my words... but found in reuben lee's transformer book....
Yes, it is true, but only under the assumption that the plate voltage is fixed and low enough that there is no danger of exceeding the plate dissipation limit.
Historically, the first need was for power gain, and the optimum load was RL=rp. This gives the maximum power for a given input (grid) voltage, assuming the grid voltage is small. This is still handy if you want to make a one-tube amplifier...
As tubes got more affordable, the main limitation became plate voltage - batteries were still common, and capacitors were expensive and of poor quality. In that case the maximum power occurs when RL=2rp. Example: a 45 at 180v. Plate resistance is 1650 ohms, so the load would be 3300 ohms and the current about 30mA (RCA says 2700 ohms and 31mA). Reich gives a good derivation of this. Notice that the plate dissipation is 5.6 watts in this case, well below what the tube can stand.
Eventually, power tubes with low plate resistance were developed, and the plate dissipation became the limiting factor. (Very old tube manuals often do not list a maximum plate power dissipation!) I've done the analysis myself because I've not found it except in fragmentary form in the usual books. Maximum power occurs when the plate voltage is maximum and the current is adjusted to result in maximum dissipation. In that case (and with a few other assumptions thrown in!) the load resistance is approximately RL=(Eb/Ib)-2.4rp. Example: a 2A3 running 300v (max rated) at 50mA (to make 15 watts plate dissipation), the load would be (300/.050)-2.4*800 or 4080 ohms.
Can you tell this is my favorite topic? :^)
RL=(Eb/Ib)-2.4rp. Example: a 2A3 running 300v (max rated) at 50mA (to make 15
watts plate dissipation), the load would be (300/.050)-2.4*800 or 4080 ohms.
could you give a bit more info on what that 4080 is????
i guess my question is, if we look at the 4080 load as the reflected load and assume a 1K signal (to keep the inductance out of the picture) i see the tube putting out the max undistoretd signal as the curves go, but when we consider the turns ratio, we will have more of a stepdown so the voltage output at the load will be less than a turns ratio that presents a 2000 ohm load... i understand the 2K load will cause the loadline to rotate... possibly making the AC swing available less so we may actually end up with less overall power even given the smaller turns ratio, but i was wondering if anyone knows if this is the case???
so by reducing the turns ratio by 30% we will get 30% more voltage, but will the reflected load presented to the tube change the tubes power out by more or less than 30%???
again its just another one of these things that needs lots of asteriks to qualify the speciffic situation.
Knowing Dave for a compleat eXperimenter, I'll just mention that everything from half this value to twice this value has been used successfully in practice by someone. Lower load impedances give more power and more distortion, gradually shading into clipping due to cutoff. Higher load impedance gives less power due to grid-current clipping, but lower distortion below clipping.
I made a more complete presentation at VSAC last month. Sometime in the next few weeks I'll get the presentation cleaned up and Brad says he'll post it and/or put it on the DVD of the seminars. That will probably be the best way to explain the assumptions behind the 4080 value. Otherwise it would take a couple napkins to write on, and a few beers!
Hey Dave, I believe you are correct, but wouldn't half the power in this instabce be fed right back through the power supply? I think max power would be with an infinite plate load and a primary load equal to 2xRp....
In Reply to: Re: Resistor as parafeed plate load? posted by JeffreyJ on October 30, 2001 at 13:55:17:
ooopps you are correct sir!!! (sound of hand smacking forehead) i wan't looking at the whole picture...
forgot about that whole voltage divider thing going on with parafeed... so while the tube would be putting out full power,
the output trannie would only see 1/2 of that power and the rest would go through the power supply...
who cares about the tube putting out max undistorted power when 1/2 of that is tossed out the window???
in this case i suppose max "useable audio" power would be from an infinate plate load and an output trannie that presents 2X the Rp...
thanks for the correction.
I thought it was where Rl = infinity. But I am unsure.
yup you are right .. see above... my bad... gave the correct info but the wrong application... i hate when i do that :-(
That was another thing I was wondering about, although I forgot to stick that in my first post. So with my output, I'd need about a 7.5K resistor to stay above a 2.5K total load on the 2A3, at which point things would get bad wrt voltage dropped and power dissipated.
However, I've been doing some more reading and found Gary P's post on using a pentode CCS on the output tube. I have the parts to try that... The only thing that I'm not sure of is how much voltage I need to have across the CCS. Say I want my the plate-cathode voltage to swing 100V each way from the static condition voltage. Do I just need 100V plus some reserve across the CCS to cover when the tube needs a higher voltage than static conditions?
The rule of thumb I've been using on the output stage is... Set the voltage on the pentode CCS equal to the plate-cathode voltage of the output tube. This gives you the compliance you need plus some margin.
On my 45 amp I have 180V across the cathode resistor to setup DC coupling, 275 Plate to Cathode on the 45, 275 across the 6bq5 pentode CCS.
From my C4S page:
"How much B+ do we need to run a Constant Current Source (CCS) on a tube? An choke fed output tube can swing above B+ to make output power. A CCS can not. So we have to plan for a higher B+ to bias a tube with a CCS. Also, we should plan on having the plate to cathode voltage of the output tube move about 10% because the plate is current fed instead of voltage fed when determining the bias point.
Version 1 (my favorite)
B+ = ( 115 V AC/ 104 V AC ) * [ V_cathode_bias + V_plate_to_cathode + I_bias * R_plate_load + Plate_margin + CCS margin]
B+ = ( 115 V AC/ 104 V AC ) * [ V_cathode_bias + V_plate_to_cathode + Max_V_swing + Plate_margin + CCS margin]
Note: Max_V_swing is the voltage swing from the plate curves between the bias point and the plate voltage with zero volts on the grid.
The standard Bottlehead C4S hard clips at about 3V across it. I've been playing with the MJE3xx models and curves. I've noticed that the (each) MJE3xx likes to see 10-20V instantaneous minimum across it instead of just 3V.
For an example, I'm going to use a 2A3 at 50 mA with about 15W plate dissipation feeding a 5K load. This CCS has three series pairs of MJE3xx transistors.
B+ = ( 110 %) * [ 60V + 310V + 50 mA * 5K + 10% of 310V for plate margin + 3 * 15V CCS margin ]
B+ = ( 110 %) * [ 60V + 310V + 250 V + 31V + 45V ]
B+ = ( 110 %) * [ 696 V ] = 766 V B+
B+ = ( 110 % ) * [ 60V + 310V + (310V - 80V) + 31V plate margin + 3 * 15V CCS margin]
B+ = ( 110 % ) * [ 60V + 310V + 230 V + 31V + 45 V]
B+ = ( 110 % ) * [ 676 V ] = 744 V B+
If we were building a standard Parafeed, B+ would have been about 382V. (60 V cathode + 310 V plate to cathode + 12 V drop in the choke) "
OK...Dave Slagle says there are no dumb ideas, so here's what I have been pondering. What about a VR tube (or paralleled pair) in place of the resistor or plate choke? I know you are looking for a slightly greater voltage drop than say a VR150 will give you and I'm betting there will absolutely no inductance-that is what we want/need in parafeed right? How about doing this in the driver circuit or in a linestage? Keeping in the same vein, how about stacking rectifier tubes to drop voltage? I have been playing with a 5Y3 in place of the cathode resistor/cap combo in a 2A3 amp. Again, I don't think you will have any inductance, but do you get any from a resistor?
i think you may be confusing a few different ideas...
for parafeed on the plate of the tube we want infiniate inductance in the plate so the VR tubes probably are not ideal as a load.
but we could look at VR tubes in the cathode to bias the tubes but this is a different issue than parafeed and will work similarly for both parafeed and direct feed... there are issues with doing this, but i have never tried it so i don't know if any of the "issues" are really issues..
now onto the good stuff...:-)
with that 5Y3... there are a couple of cool things you could look into...
first, be a man and use a 280 :-) looks count for something.
seriously, what voltage drop are you getting the book says 60V @ 125ma so run 70ma and you might be close to -40V???
also you can use one tube for a pair of outputs, somehow there has to be beauty to that approach...
and finally... try tweeking the filament voltage to adjust the bias... you should be able to trim the filament voltage by 10% or so and get pretty close to the book points for the 2a3, and the rectifier should still be operating linearly... it might be a very easy way to build a simple amp??? also using an indirectly heated rectifier will give you a nice slow turnon... but as it looks now the 5Y3 80 5Z3 family looks best someday i'll scour the books a bit more and see what turns up...
also how about an 83 to bias up a 12B4 linestage for a nice fixed 10-15V bias??? a friend said some industrial applications used mercury tubes for bias due to the huge currents, a resistive load would dissapate too much heat..
just food for thought.
Forgot to mention that I'm running a traditional SE OPT amp so the parafeed inductance issue doesn't apply for me, I just tossed it out as a thought.
Ahhh...the 280! Hard to come by in these parts because all of the old radio restorers seem to hoard em. Found 1 at the last Hamfest but it resides in one of my linestages...you are right, it's a great looking tube! I had not even though of that one for this application. I do have some NOS 80s that I'll have to try but do you think I might be pushing the current limits using a single tube for both 2A3s? Series 5AR4s seems kind of excessive in my little budget amp and I'm not sure the voltage drop would be right.
I don't have enough B+ to get direct coupled on this amp. That led me to using a rectifier tube on the cathodes instead of a VR type. The 5Y3 drops 35 volts for 115ma for the pair of 2A3s. Neg leg of the 5vDC supply gets grounded, right...that accounts for the reduced hum on the output?? Sounds very fast! Now I have to play with the driver stage and coupling caps to get a little more low end. It's nice and tight but I would like a touch more of it.
But how about a VR tube from B+ to plate in a cap coupled output or WOT linestage...or in the driver section of an amp?
i'm confused here...
are you using a single 5Y3 per tube, or are you connecting both plates to the respective cathodes of the 2a3's and the filament of the 5Y3 to ground???
you say 35V @ 115ma, but how are you getting 115ma??? i am asuming its the combined current through both tubes, but you have to consider the current for each rectifier plate, which is just under 60ma and a 35V drop which sounds about right...
you also mention a 5VDC filament... is that for the 5Y3??? did you try ac... it just seems odd rectifying the filaments of a rectifier... just curious... i never considered the filament as cathode of the directly heated rectifiers... i suppods that could be an issue.
Sorry, too many variables without drawing up a schematic.
Yes-1 5Y3 but with the plates tied together. Plates to the CT of the common 5v AC filament tranny for both 2A3s. I tried AC to heat the 5y3 but it was a 60Hz hum machine! I thought...try a DC supply to heat the 5Y3, you have a big ol' Lambda taking up space that puts out tons of current. Voila! No cap bypassing the 5Y3 at that point. Lacking bass but otherwise a great improvement over a cathode resistor. I even switched back to hear if I was just thinking the rectified cathode was really better....yep, it was...errr...is, at least to my ears.
I have been playing with different caps bypassing the 5Y3 to bring up the low end. Maybe this "shouldn't" be done, but I'm just in a mood to try anything that sounds feasable. There really isn't anything expensive in this amp if it goes up in flames...except the OPTs. But they are really in little danger here...the main rectifier(an old EZ80) will fail before I get anywhere close to the current rating of the OPTs. This has been an interesting experiment...and it's sounding better!
Any thoughts or advice? I'm all ears tonight.
not too many variables... i just wanted to be clear...
since the plates are tued together, i would think you would be getting the equivalent of 30ma of current per 1/2 since they
are in parallel, you could try just using 1 plate and your bias should go up because you will be drawing double the current...
its worth a try.
i never thought about the 60hz hum when using an DHR... so thanks for that pointer... maybe thats the problem with the 83
how about tryong a hum-balance on the rectifer??? i think its probably step in the wrong direction, but it would be
interesting to try i would use gordons trick of a 100 ohm pot bypassed with 10 ohm resistors and maybe bypass them with
you could also try bypassing just the filament with caps... again wrong direction i suspect but something may be found
i don't much care for the filament supply for the rectifier... so maybe the search must start for that elusive inefficient
IDR.... damn.... i opened up the book and out jumped the 6853... looks like an IDR same specs as the 5y3 but it shows a cathode (filament is connected to the cathode) ...probably as easy to
find as the 8212 and the 8213...
also the 6106 looks like a possibility, but they are listed (by AES) as a 5Y3-wgt... so you would have to be sure you got the right one... but they may be easier to find...
I was talking to one of our older engineers (who probably thinks I'm nuts..."just use a resistor") and he suggested a hum pot also. He thinks that with a good quality CT 5v tranny, there might not even be need for it. The tranny I was using had no CT and that's why I simply grounded on side and had all that hum.
Gotta run again,
ahhh that changes things a bit :-)
another thing to try is a pair of say 5 ohm WW resistors in series across the winding and ground the "virtual" CT... you could even size theseresistors to load down the trannie a bit and drop your filament voltage below 5V and get a larger bias???
lots of things to explore here.. have fun!
I used an 866A under the cathode in a preamp I was fiddling with a few weeks ago. It looked really cool, but there were problems with noise. I'm not sure if it was from that or something else - it was a new design that I just sort of did for kicks and I've disassembled it since. It only dropped about 10 volts, though.
was it the same 60hz noise tom mentioned from the 5Y3???
Well, there was 60hz hum. Plenty of it. However, I was using SV572-10's as a linestage, so that could have been the cause of that. The reason I lost interest was that there was a ton of higher frequency hash. When I turned on my speakers, there was just solid noise. I don't have a scope, so I couldn't look at. I didn't really ever track down what the problem was. I just unclipped it all and moved on to other things. It looked cool, though - the bright white 572-10's and the blue 866a's (I had them in the supply too).
Reading your other posts now, I did try a hum balance pot on the 866a filament - it helped some, but not enough. I had a huge amount of voltage present on the outputs initially and tweaking the hum pot got it low enough for me to actually turn on my speakers to listen to. Well, I never actually listened to anything through it...
This post is made possible by the generous support of people like you and our sponsors: