@Wasnex doing the good work I was too lazy/busy to do.
I don't even own an SWR gear, but I was curious and got what appears to be the perfect "hit" on my first Google search
. Then again, I wouldn't mind owning an IOD some day
.
Enjoy the benefits of a premium membership!
Upgrade Account
-
TalkBass has been independent since 1998. Add your voice.
Create a free account to reply to discussions, view embedded media, and browse with fewer display ads.
Join free Log in
Want zero display ads or expanded classifieds tools? Compare plans.
SWR Interstellar OD - Retubed
- Thread starter Low Down Brown
- Start date
I don't even own an SWR gear, but I was curious and got what appears to be the perfect "hit" on my first Google search
. Then again, I wouldn't mind owning an IOD some day
.
Thank you, but as I indicated in my first post, I already tried the AE Mod and didn't like it. I was just looking for the IOD biasing instructions. Those are the instructions I was going to be studying before attempting. But nonetheless, thank you for your efforts.
Sorry, I got the impression others were interested in the bypass mod. Sorry if I misread. The bias procedure is included in post #17.
If your feel the bypass mod I linked is too extreme, consider using a smaller value capacitor. The original capacitor is 0.001 and the value for flat is 0.1, so there is a huge range of values that can be tried that would provide a less scooped response.
I'm definitely not an expert on SWR or tube gear, but I talked to 2 different amp techs during the course of researching and retubing the IOD. I can't technically explain why, but both said that it wasn't necessary to bias the IOD after putting in new tubes. So, I didn't mess with mine and it's absolutely fine. I think it was something along the lines of "yeah, it has power tubes, but they're not actually performing as power tubes, so....No big deal."
Something like that...
Something like that...
Per the manual the output tubes are setup to produce 5W RMS. So they are actually working as power tubes, although the output power suggests they are run very conservatively. I think a normal amp with a pair of EL34s would probably make at least 20W.
One thing that concerns me is there are no screen resistors, and the screens are fed from the same supply as the center tap of the output transformer (OT). This means the screens will actually idle at a slightly higher voltage than the plates, when they should be a bit lower voltage. This is because the OT will drop a few volts from the center tap to the outer windings.
Pushing the amp into hard overdrive will magnify this problem because the OT will drop even more voltage when the tubes are working hard and pulling a lot of current. A ramification of this is the bigger the voltage differential between the screens and plates, the more the screens will try to hog the current, which can push the screens beyond their dissipation limit. Per tube data sheets, the plate dissipation limit is 12W and the screen limit is 2W, so you can see how it would be a problem for the screens to try and act like they are the plates.
The provided bias guides recommends setting the bias no higher than 70% dissipation, or 8.4W. This is a common limit for AB1 push pull amps; however, due to the fact there are no screen resistors, I have to wonder if biasing at 70% is actually too aggressive. If it were my amp, I would see how it sounds over a range of 40-60%. However since it appears the tube is being run fairly conservatively, it may be possible that any range of the bias control is fine.
I would modify the recommendation in the bias to guide to include checking for red plating and glowing screens when the amp is pushed into hard overdrive.
First of all, a happy Summer Solstice to everyone!
Summer arrived, as it often does here, with somewhat cool and damp weather. What better way to warm it up than with some cheery, glowing vacuum tubes? I've been meaning to work on my Stella and this was the perfect excuse.
I've been thinking about the instructions for setting the bias (earlier post in this thread). 70% plate dissipation is a good target for typical class AB1 tube power amps, but Stella is class A, does that really apply? So I took off the lid and got out my DVM.
My first step was to measure the DC resistance of the output transformer primary. Knowing that resistance and the B+ voltage at the screens, measuring the plate voltage and dividing by the primary resistance gives the plate current. No need to mess around with the current measurement as described in the bias instructions. (I blew the fuse in my old DVM by carelessly touching a probe to the wrong place the last time I tried that - dang fuse is unobtanium!)
It turns out the resistances between the center tap and either side of the primary are different! I measured 292 ohms from brown (center tap) to the black lead, 368 ohms to the red lead. I suspect the transformer isn't bifilar wound (where the two sides of the primary are created by winding two wires side by side). Instead, a given number of turns are wound for one side of the primary, then the same number on top of that for the other. The second winding is wound over the top of the first - so the same number of windings ends up being more wire and thus higher DC resistance.
In addition to the resistance of the output transformer primary, there's a 270 ohm resistor in the power supply. The result is that as the bias is adjusted to be less negative and the plate current increases, the plate voltage drops. The end result is that it's really not possible (at least, with the JJ EL84s I have in there now) to push the plate dissipation above about 70% of the 12 W maximum. This probably also helps save the screens, since the screen voltage drops somewhat along with the plate voltage due to the voltage drop across the 270 ohms.
This is actually kind of good news. It means that it's really hard to hurt anything with the bias adjustment, so adjusting it for sound rather than plate dissipation is almost foolproof. (You could probably overheat the power transformer if the bias is too hot, but I expect it would sound like @@@ at that setting.) I went ahead and started tweaking it by ear, and found that I liked it best biased much cooler. I settled on something around 4.25 W, about 35%, with a bias of -10.5 V and B+ 267 V.
Another thing to consider: the schematic shows B+ as 250 V. Adjusting for that gave about 5.8 W which is 48%, and sounded about the same.
So maybe the right way to adjust bias is easy: set B+ to 250 V. as a starting point and adjust from there, listening for a sound you like.
Biasing cooler will lead to a less abrupt onset of distortion - it will tend to drive one tube to approach cutoff (soft clipping) before the other tube's grid tries to go positive (which will probably clip hard, I don't know if the transistor (!) driver can push enough current to drive the grid positive).
I also noticed that my tubes were not very well matched - the one on the higher-resistance side of the output transformer, and thus with lower plate voltage, was actually drawing more current. I swapped them, and the idle dissipation was much better balanced. Maybe it's time to try a pair of those TungSols.
A couple other points:
You meant EL84s, I think
Summer arrived, as it often does here, with somewhat cool and damp weather. What better way to warm it up than with some cheery, glowing vacuum tubes? I've been meaning to work on my Stella and this was the perfect excuse.
I've been thinking about the instructions for setting the bias (earlier post in this thread). 70% plate dissipation is a good target for typical class AB1 tube power amps, but Stella is class A, does that really apply? So I took off the lid and got out my DVM.
My first step was to measure the DC resistance of the output transformer primary. Knowing that resistance and the B+ voltage at the screens, measuring the plate voltage and dividing by the primary resistance gives the plate current. No need to mess around with the current measurement as described in the bias instructions. (I blew the fuse in my old DVM by carelessly touching a probe to the wrong place the last time I tried that - dang fuse is unobtanium!)
It turns out the resistances between the center tap and either side of the primary are different! I measured 292 ohms from brown (center tap) to the black lead, 368 ohms to the red lead. I suspect the transformer isn't bifilar wound (where the two sides of the primary are created by winding two wires side by side). Instead, a given number of turns are wound for one side of the primary, then the same number on top of that for the other. The second winding is wound over the top of the first - so the same number of windings ends up being more wire and thus higher DC resistance.
In addition to the resistance of the output transformer primary, there's a 270 ohm resistor in the power supply. The result is that as the bias is adjusted to be less negative and the plate current increases, the plate voltage drops. The end result is that it's really not possible (at least, with the JJ EL84s I have in there now) to push the plate dissipation above about 70% of the 12 W maximum. This probably also helps save the screens, since the screen voltage drops somewhat along with the plate voltage due to the voltage drop across the 270 ohms.
This is actually kind of good news. It means that it's really hard to hurt anything with the bias adjustment, so adjusting it for sound rather than plate dissipation is almost foolproof. (You could probably overheat the power transformer if the bias is too hot, but I expect it would sound like @@@ at that setting.) I went ahead and started tweaking it by ear, and found that I liked it best biased much cooler. I settled on something around 4.25 W, about 35%, with a bias of -10.5 V and B+ 267 V.
Another thing to consider: the schematic shows B+ as 250 V. Adjusting for that gave about 5.8 W which is 48%, and sounded about the same.
So maybe the right way to adjust bias is easy: set B+ to 250 V. as a starting point and adjust from there, listening for a sound you like.
Biasing cooler will lead to a less abrupt onset of distortion - it will tend to drive one tube to approach cutoff (soft clipping) before the other tube's grid tries to go positive (which will probably clip hard, I don't know if the transistor (!) driver can push enough current to drive the grid positive).
I also noticed that my tubes were not very well matched - the one on the higher-resistance side of the output transformer, and thus with lower plate voltage, was actually drawing more current. I swapped them, and the idle dissipation was much better balanced. Maybe it's time to try a pair of those TungSols.
A couple other points:
With more plate voltage, and 70% dissipation, I think you're right. That might be into class AB territory, though.
You meant EL84s, I think
I think you are onto something here; what I measured tends to confirm this. The highest plate dissipation I was able to achieve was at only about 180V on the plate and 196 V. on the screen, and I think it sounds better run much cooler than this.
First of all, a happy Summer Solstice to everyone!
Summer arrived, as it often does here, with somewhat cool and damp weather. What better way to warm it up than with some cheery, glowing vacuum tubes? I've been meaning to work on my Stella and this was the perfect excuse.
I've been thinking about the instructions for setting the bias (earlier post in this thread). 70% plate dissipation is a good target for typical class AB1 tube power amps, but Stella is class A, does that really apply? So I took off the lid and got out my DVM.
My first step was to measure the DC resistance of the output transformer primary. Knowing that resistance and the B+ voltage at the screens, measuring the plate voltage and dividing by the primary resistance gives the plate current. No need to mess around with the current measurement as described in the bias instructions. (I blew the fuse in my old DVM by carelessly touching a probe to the wrong place the last time I tried that - dang fuse is unobtanium!)
It turns out the resistances between the center tap and either side of the primary are different! I measured 292 ohms from brown (center tap) to the black lead, 368 ohms to the red lead. I suspect the transformer isn't bifilar wound (where the two sides of the primary are created by winding two wires side by side). Instead, a given number of turns are wound for one side of the primary, then the same number on top of that for the other. The second winding is wound over the top of the first - so the same number of windings ends up being more wire and thus higher DC resistance.
In addition to the resistance of the output transformer primary, there's a 270 ohm resistor in the power supply. The result is that as the bias is adjusted to be less negative and the plate current increases, the plate voltage drops. The end result is that it's really not possible (at least, with the JJ EL84s I have in there now) to push the plate dissipation above about 70% of the 12 W maximum. This probably also helps save the screens, since the screen voltage drops somewhat along with the plate voltage due to the voltage drop across the 270 ohms.
This is actually kind of good news. It means that it's really hard to hurt anything with the bias adjustment, so adjusting it for sound rather than plate dissipation is almost foolproof. (You could probably overheat the power transformer if the bias is too hot, but I expect it would sound like @@@ at that setting.) I went ahead and started tweaking it by ear, and found that I liked it best biased much cooler. I settled on something around 4.25 W, about 35%, with a bias of -10.5 V and B+ 267 V.
Another thing to consider: the schematic shows B+ as 250 V. Adjusting for that gave about 5.8 W which is 48%, and sounded about the same.
So maybe the right way to adjust bias is easy: set B+ to 250 V. as a starting point and adjust from there, listening for a sound you like.
Biasing cooler will lead to a less abrupt onset of distortion - it will tend to drive one tube to approach cutoff (soft clipping) before the other tube's grid tries to go positive (which will probably clip hard, I don't know if the transistor (!) driver can push enough current to drive the grid positive).
I also noticed that my tubes were not very well matched - the one on the higher-resistance side of the output transformer, and thus with lower plate voltage, was actually drawing more current. I swapped them, and the idle dissipation was much better balanced. Maybe it's time to try a pair of those TungSols.
A couple other points:
With more plate voltage, and 70% dissipation, I think you're right. That might be into class AB territory, though.
You meant EL84s, I think
I think you are onto something here; what I measured tends to confirm this. The highest plate dissipation I was able to achieve was at only about 180V on the plate and 196 V. on the screen, and I think it sounds better run much cooler than this.
Very cool!
Why do you say Stella is Class A? This is from the owner's manual: "We tried class A type power amps and push-pull. Push-pull was decided upon as it seemed more musical and even." So I don't think SWR considered the design class A.
Yes I meant EL84s...oops!
The bias guide is a bit confounding. One sections say 70% dissipation or 8.4W is max. The same section also essentially says the hottest tube should not exceed 7W, which would be about 58.3333% dissipation. Another section says <7W is conservative dissipation. The computed example in this sections comes out to 5.7W, which is described as "in range." 5.7W would be 47.5% dissipation.
I have seen others that say it's impossible to bias the IOD hot enough to cause damage (I don't now if this is true). I do think at the very least it would shorten tube life.
Here what I think. The hotter you bias the tubes, the more the screen voltage will exceed the plate voltage in this particular design, and I can't imagine that is a good thing. I believe the screens have a 2W limit while the plates have a 12W limit. When the screen voltage exceeds the plate voltage, AFAIK the screens tend to hog the current, which can push them over their dissipation rating.
So if the IOD sounds best biased fairly cold
.I think that's an unclear statement in the owner's manual. I'm pretty sure what they meant was "We tried class A single-ended type power amps and push-pull."
Class A and push-pull are not opposites, they are two completely different sorts of characteristics. Class A simply means that the output device or devices conduct through the full cycle of the waveform. It's necessary for single ended audio amplifiers, unless you're deliberately seeking significant distortion. It's also possible to operate push-pull amplifiers in class A; this is not uncommon in hi-fi tube power amps (like my Heathkit W5M, which makes an awesome bass amp, by the way).
Class A is inefficient, class AB push-pull improves efficiency by biasing the output devices closer to their cutoff voltage. This shuts off one output tube for part of the cycle while the other one works harder. This would introduce significant distortion, so negative feedback has to be used to tell the tube that's on that it has to work harder for a while. This is usually done by a connection (a resistor and capacitor in parallel) from the secondary of the output transformer back to the cathode of a voltage gain stage earlier in the power amp.
The little push-pull power amp in the Interstellar Overdrive doesn't have this feedback loop, so I see it as operating in class A push-pull when it's not being driven hard. Under this condition, it gives that nice tubey warmth without overt distortion. Of course, it has "overdrive" in the name, so, depending on how it's biased it may end up driving the output tubes into cutoff (class AB, but without feedback to limit the distortion), or it may try to drive the output tube grids positive. That starts drawing grid current, which will start charging up the coupling capacitors, or may simply overload the phase splitter (a transistor, sigh).
I think it sounds better with the soft clipping as the output tubes are driven towards cutoff than with the hard clipping of trying to drive the grids positive. My JJ EL84 datasheet says it takes about -15 volts on the grid to cut it off. Adjusting bias so the B+ is dragged down to 250 V like it says on the schematic results in bias around -9 V (for my tubes, yours may differ). So it's about 6 V away from cutoff, and 9 V away from positive grid territory.
I know, another long-winded post. But that's why I call it class A.
Class A and push-pull are not opposites, they are two completely different sorts of characteristics. Class A simply means that the output device or devices conduct through the full cycle of the waveform. It's necessary for single ended audio amplifiers, unless you're deliberately seeking significant distortion. It's also possible to operate push-pull amplifiers in class A; this is not uncommon in hi-fi tube power amps (like my Heathkit W5M, which makes an awesome bass amp, by the way).
Class A is inefficient, class AB push-pull improves efficiency by biasing the output devices closer to their cutoff voltage. This shuts off one output tube for part of the cycle while the other one works harder. This would introduce significant distortion, so negative feedback has to be used to tell the tube that's on that it has to work harder for a while. This is usually done by a connection (a resistor and capacitor in parallel) from the secondary of the output transformer back to the cathode of a voltage gain stage earlier in the power amp.
The little push-pull power amp in the Interstellar Overdrive doesn't have this feedback loop, so I see it as operating in class A push-pull when it's not being driven hard. Under this condition, it gives that nice tubey warmth without overt distortion. Of course, it has "overdrive" in the name, so, depending on how it's biased it may end up driving the output tubes into cutoff (class AB, but without feedback to limit the distortion), or it may try to drive the output tube grids positive. That starts drawing grid current, which will start charging up the coupling capacitors, or may simply overload the phase splitter (a transistor, sigh).
I think it sounds better with the soft clipping as the output tubes are driven towards cutoff than with the hard clipping of trying to drive the grids positive. My JJ EL84 datasheet says it takes about -15 volts on the grid to cut it off. Adjusting bias so the B+ is dragged down to 250 V like it says on the schematic results in bias around -9 V (for my tubes, yours may differ). So it's about 6 V away from cutoff, and 9 V away from positive grid territory.
I know, another long-winded post. But that's why I call it class A.
I don't think you can exceed the maximum plate dissipation in the tubes at any setting of bias, but my rough estimate says you could dissipate around 10-15 watts in the output transformer primary and more than that in the 5W 270 ohm resistor in the power supply. So biasing it too hot would be bad.
Very possible. Also, I would be concerned with exceeding the 2W dissipation rating of the EL84 screens.
I believe at least part of the definition is how you bias the output section.
From The Last Word on Class A
How is class AB defined?
A class AB amplifier is one in which the grid bias is set so that plate current flows for more than half, but appreciably less that the full 360 degrees of the the input cycle, again measured at the full, unclipped output of the amplifier.
Since the max dissipation for the IOD is 70% and <58.3% is recommended, it seems to conform more to AB.
I suspect it would operate as class A at very low volume.
I don't know whether to thank you or curse you for that link - I'm sure I'll spend many hours reading there!
The article had a couple things that really stood out to me. First: "because a class AB amplifier is biased so that the plate current flows for the entire cycle at lower output levels (which is done to reduce crossover distortion), many people claim it is a 'class A amplifier at lower volumes'. This is simply not true." This summarizes where I went wrong calling the Interstellar Overdrive "class A."
The other interesting thing was: "The non-negative feedback amp transitions much more smoothly into distortion, making it better for players who like to use their volume control to change from a clean to a distorted tone." That's probably part of what makes the Interstellar Overdrive work so well, and also probably why I like it better biased for lower plate dissipation.
Now I need to go back and read the article on speaker load emulation....
The article had a couple things that really stood out to me. First: "because a class AB amplifier is biased so that the plate current flows for the entire cycle at lower output levels (which is done to reduce crossover distortion), many people claim it is a 'class A amplifier at lower volumes'. This is simply not true." This summarizes where I went wrong calling the Interstellar Overdrive "class A."
The other interesting thing was: "The non-negative feedback amp transitions much more smoothly into distortion, making it better for players who like to use their volume control to change from a clean to a distorted tone." That's probably part of what makes the Interstellar Overdrive work so well, and also probably why I like it better biased for lower plate dissipation.
Now I need to go back and read the article on speaker load emulation....
. 
Haha interesting! Not sure how the EQ would work with an instrument level signal. Only one way to find out.
Oh also, the Master Volume control is also after the power amp, so cranking the Master, like I've been doing for years, doesn't actually run the tubes any hotter. There does seem to be some audible effect on the tone as you turn the master up, but it's not coming from the tubes.
