• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Trouble with a paraphase PI

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hey, I recently converted a Bogen K15 into what I thought would be a fairly nice guitar amp. I redid the preamp but the PI was a paraphase type, I'm not too familiar with them and there's not a lot of info out there about them so I just kind of left it the ay it was.

Anyway, it didn't turn out as good as I thought, it sounds great clean but when you play "on the edge" of breakup, you can hear some nasty distortion mixed in. I ran through it with my scope and I saw that the source of the nastiness was indeed the phase splitter stage. There are two preamp stages before it, after the first one there's no breakup at all (as expected) and after the second there's a little smooth, asymmetrical clipping, also as expected. However, when I checked the signal coming from the "first" half of the PI, I noticed that it would reach a certain pont and clip hard, very SS like, and only one side of the wave. This signal goes to one output tube and also the grid of the other half of the PI. When it gets here it "flips" it (as it should) and does the same thing to it, so now both sides of the wave are clipping hard, and this goes to the other output tube. This, as you can imagine, sounds like a mess.

This is the original K15 schematic by the way. Don't bother looking at the preamp, just the two halves of V3 onward.

http://www.flickerdown.com/forsale/k-15.pdf

So waht I want to know is, is this "SS clipping" inherent with this style of phase splitter? I tried a few different tubes (a couple Sylvanias and a Mullard) and they all did the same thing. Can I tweak some values to get it to sound better or am I going to have to switch to a LTP design? I really don't want to do that because this amp is a horribly laid-out mess inside (not my fault, it was a mess from the factory) and I'd rather not go to that extent with this amp unless it's 100% neccesary. I suppose there's a reason why everyone stopped using this design in the 50's though.

The original components are still in the PI stage. What exactly constitutes a "leaky" cap? I checked the coupling caps and they were only letting a few mV get through, is that enough to cause a problem?

Thanks in advance.

-Darren
 
Darren,

Letting a few mV. thru is leaking, not badly, but still leaking. Replace the coupling caps. with Orange Drops.

The paraphase splitter is out of favor. Things like that usually happen for a reason. What you observe is likely 1 of a number of reasons.

The big advantage of paraphase topology is gain. Less gain is available when differential splitter topology is used. Let's see, if a 12AT7 differential splitter replaces the 12AX7 paraphase splitter, some of the lost gain could be made up by buffering the splitter triodes with DC coupled MOSFET source followers as described in MOSFET Follies. An advantage of differential splitter topology is the very convienent point the non-inverting triode's grid provides for the application of loop NFB. You get a short loop around only 2 stages.

BTW, 200-220 V. on the 'T7 plates and IB = 3.5 mA. sounds good.
 
No, that kind of clipping is not a characteristic of paraphase inverters. Most likely it is a consequence of NFB. Just for kicks, you might try changing the NFB connection from the 70V tap to the 25V or 16 ohm tap. The wire is probably soldered right to the speaker terminal on the back of the chassis, which might not make it easy to to get to...

I have used the paraphase inverter and found that it can sound quite good, both for creating music and recreating it. My suspicion is that it fell out of favor because (a) it does not look as elegant on paper as a concertina or LTP, and (b) because it makes it more difficult to apply NFB without squirrely behavior. Anyway, I once built a guitar amp using a paraphase inverter and I thought it sounded very good. It had a nice transition from clean to overdrive that was easy to control and smooth. Just the opposite of what you describe.

As for the coupling caps, a few mV is normal and is the result of a minuscule amount of grid current. Are the grids a few mV positive or negative? Positive means leaky caps, negative means grid current which is harmless.

-- Dave
 
Did you try replacing the 12AX7 with a lower gain 12AU7?

Did you measure the voltages at pin 6, which should be 155-160V. That should leave enough swing before clipping at 300 or 1.5. (Is the clipping occurring at the top or bottom?)

What did you do for the preamp stage? The 12AX7 and 6AV6 may have had enough gain for guitar.

Also check pin 8 for 1.5V. If it's off, check C12.
 
Ex-Moderator
Joined 2004
I agree that a paraphase splitter could be made to sound OK in a guitar amp, but I'm surprised to see 12AX7 driving 6L6s. I'd expect the splitter to be subject to overload. I think you might be better off with a more rugged tube like 12AT7 or even the (much maligned) 12au7. You'd need to reduce the splitter's plate load resistors from 220k to around 47k and reduce its cathode resistors from 2.2k to around 470Ù. The gain would be reduced singificantly. Alternatively, Eli's idea of buffering the 12AX7 splitter with MOSFETs (e.g. IRF820) could work well, save you from losing gain and give better overdrive behavior of the 6L6s.
 
I'm surprised they used such high plate resistors on the 12AX7 - with 220K load and 470K grid on the next stage. Going to 100K (with cathode resistors halved as well) would give a little more drive.

As already suggested, the feedback may be the problem - remove it (150K resistor) and see how it sounds - should reduce the abrupt transition to clipping.
 
Eli, I'll replace 'em with orange drops then. I don't think I have any laying around of that value but I'll grab some soon. I know the orange drops I used in the preamp don't leak at all.

Dave, I actually wired an SPST switch in the NFB loop, one of the first mods I did actually. I never thought to shut the NFB off when I was testing it on the bench though, I typically keep it on unless I really want to crank it and want the extra distortion. I'll shut it off and check again but if I remember correctly I did hear pretty much the same thing when it was turned off. The voltage I was getting was positive. I guess it's for sure that the old ones need to go then.

PRNDL, I remember checking voltages but I can't remember off hand what pin 6 was. I think it was somewhere in that area though. I'll go check again. Clipping is happening at the top. I'll try a 12AU7 in there tomorrow and see what happens, I wish I had a 12AT7 to try as well. I know the preamp is fine because it's pretty much identical to the one I used in my CHB-10A. Kind of similar to a 5D2 princeton but with no tone control. That preamp driving a single 7868 is the greatest guitar amp I have ever played through. I used the same preamp in this amp hoping to get the same results but the darned PI is in the way :)

Ray, I believe there was some Fender models that had a 12AX7 driving 6L6s early on. I know the 5D5 Pro did for sure. My amp actually has Tung-Sol 5881s in it right now, once they start to break up it sounds great, but I can still kind of hear the phase splitter's nastiness even when it's dimed.

Tom, I was actually thinking of trying to duplicate the one out of the 5D5 actually. It uses 100k I believe.

Thanks for the ideas guys, I'll check out some of this stuff when I get a chance tomorrow.

-Darren
 
I guess I'll stick to SE amps from now on :).

Anyway, I checked pin 6 and it was 158VDC, just fine. Pin 8 had 1.5, as expected.

I know the preamp is driving the PI pretty hard. To me it seems like it should be clipping smoothly though, instead of the harsh abrubt clipping that it's doing. If "too much gain" really is the problem, then I will probably either switch to an LTP design or abandon this project altogether and get started building something useful. I'm just not uderstanding why it's clipping the way it is, I don't mind some distortion coming from the PI but it can't sound like this - I've never come across this before and I've also never messed with a paraphase style before so I just kind of assumed the general design must be the reason why it sounds bad.

I'll try a 12AU7 in there the way it is now and see what happens.

Forgot to mention, I shut the NFB off and I do indeed get the same thing happening.
 
After thinking about it a little, I realized that the hard clipping is probably due to driving the output tube grids positive (DUH!) Adding 'grid stoppers' to the 6L6's will almost certainly change the overdrive characteristics of the driver. You might try 1k to 10k as an experiment. Going even larger might make the overdrive smoother (or not) but I have found that going too big makes the clean tone suffer. I don't know why, but it does.

Another after thought since my other post: I'm still not convinced that the coupling caps are leaky. RE: the few mV present on the 6L6 grids and my positive vs. negative comment, I was thinking about so-called grid leak bias that was sometimes used when only a volt or so of bias was needed. In that case the 'idle' grid current results in a small voltage drop across the grid resistor which biases the grid negative. Evidently, in power tubes there other effects (which I admit I have not completely figured out) including secondary emission of the grid and so-called gas current. These currents are in the opposite direction; they result in a positive voltage drop across the grid resistor.

That is the reason that a maximum size grid resistor is sometimes given in the manufacturer data sheets. Too large a resistor can result in runaway; grid current causes a small positive shift (less negative bias) resulting in more cathode current which results in more grid current shifting the grid voltage even more which results in even more cathode current.... Cathode bias tends to tame this effect, so the maximum grid resistance is generally larger than when fixed bias is used.

Anyway, I stick to my claim that a few mV is normal and not necessarily the result of leaky coupling caps. I do want to take back my comment about negative vs. positive.

If you do swap out the coupling caps for something better I would not recommend Orange Drops in this application. They work well in small signal applications (they're woolly to my ears, but that's just my opinion) but I think that at this signal level they suffer. Those cheap little metalized poly things are even worse. Don't let the 600VDC rating fool you. They just can't handle the AC voltage and current and still sound good. Notice that Fender sometimes used Orange Drops for preamp coupling and in the tone controls, but they used physically larger and more expensive paper caps to couple to the output grids. My own personal choice is paper/oil or poly/oil even in guitar amps.

-- Dave
 
No deal on the grid stoppers. I tried 5.6K (standard Marshall value) on each tube and it sounds bloody horrible. I'm not sure if it really changed or not, or if it just sounded worse because I was angry I wasted my time adding the resistors.

I think my only option here is a LTP 12AT7 setup or something along those lines.
 
Well, it turns out it's not a result of being driven too hard from the preamp.

I was messing around today and bypassed the second preamp stage to see what it would do. Same distortion. Frustrated, I bypassed the whole preamp and fed the guitar input straight to pin 7 of the PI tube. When I hit a few chords pretty hard, guess what I could hear? If it's behaving this way with just a low-level guitar output fed into it, too much preamp gain is clearly not the problem here.

I think I'm going to try new .022uF coupling caps and see what happens. I have some orange drops of that value but as per Dave's advice I'll get something else.
 
Hey Vinnie, I'm sorry that you wasted your time with the grid stoppers.

At this point it's pretty hard to tell what's going on, especially without actually hearing the thing. Could be a flaky component ... sometimes just cleaning the sockets and tube pins can fix a splatty distortion.

As I said, I'm out of good ideas at the moment.

-- Dave
 
Fair enough.

My usual supplier of caps is out of .033 oil/paper caps right now, but I do have the .022 orange drops sitting around. They may not be the greatest but I think they may be worth trying just to make sure the others really aren't leaking. I'll play around with some component values as well.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.