Orange OR120 with issues (odd clipping, PI balance, voltages, etc.)

[TheKman76]

To reiterate, by reducing the resistance to 100k @FleshOnGear did add some gain

I'm afraid I don't agree. A quick LTSpice model confirms for me that gain is very slightly lower with a small grid leak.

I always figured they increased Ra of the driver stage to 390k to bring the Q point closer to the proper voltage for the grid of the DC coupled cathodyne stage.

Probably, it's unusually high with the ECC83.

From a technical standpoint I feel the original design is the better design.

 

[FleshOnGear]

Eureka! I found the bad solder joints on the top reservoir cap! It took my poor eyes a moment to see it, but yeah, they were some shitty solder joints. Everything is back to normal, except I have sane voltages in the amp. HT is about 470V now. As I reported earlier, heaters are at 6.5VAC.

I had to stop to start making dinner, but a quick play test was promising. Other than no modulation, the onset of clipping sounded much more gradual. I came to this conclusion after only about 1 minute of playing, so we’ll see if I change my mind.

On the whole, I am very relieved and pleased! Thank for all the help, folks!

 

[TheKman76]

Good result, sir!

So the rest of the amp, PPIMV aside, is still stock? And the distortion is better?

 

[FleshOnGear]

Good result, sir!

So the rest of the amp, PPIMV aside, is still stock? And the distortion is better?

Right now the 110k resistor is still piggybacked on the 1M PI grid resistor. Like I said a bit ago, I’m not sure how much it really did, so I’m going to remove it tonight. Other than that and the MV, the circuit is stock.

I think there’s a sweet spot for EL34s, and it’s under 500V. Also, it’s possible that there was a little bit of resistance between the wires and the reservoir cap before I knocked them loose, and subtle modulation might have impacted the distortion tone. That’s just speculation, though. What I heard today sounded better.

 

[TheKman76]

Right now the 110k resistor is still piggybacked on the 1M PI grid resistor.

Do you think this had a greater improvement than removing the NFB? I'm really interested to know if the cathodyne is the source of the issue or not.

 

[Pete Farrington]

Eureka! I found the bad solder joints on the top reservoir cap!

Great, well done, the symptoms were a textbook case of massive HT ripple.

 

[FleshOnGear]

Do you think this had a greater improvement than removing the NFB? I'm really interested to know if the cathodyne is the source of the issue or not.

No, I don’t. And I don’t believe the cathodyne was the source of the offending distortion. At this point I believe the poor connection to the reservoir cap or the abnormally high HT were to blame. I plan to get more voltage readings of the cathodyne stage to see where it settles now, and I plan to play test a bit more, before making any conclusions. For me, I didn’t suspect the frequency doubling distortion as an issue because the errant distortion I heard went away when I turned down the PPIMV a bit.

Great, well done, the symptoms were a textbook case of massive HT ripple.

That makes sense. In this case I put too much trust in some stranger’s amp tech to wire up a filter cap properly. I need to learn to trust my gut over some other guy’s tech. Thanks for the help and proper diagnosis.

 

[FleshOnGear]

@TheKman76 sorry I keep flip flopping on this. After some more play testing, I’ve realized that reducing the PI grid resistance does make a real difference. I wasn’t imagining it. It’s not so much a difference in how gradually the distortion comes on, it’s the quality of the distortion.

When the grid resistance is lower (1Meg//110k) the onset of distortion is crunchier and more satisfying. When the grid resistance is 1Meg, the distortion is fizzier and sounds like static electricity snapping.

I think IT IS the cathodyne distorting. Earlier I thought that turning down the PPIMV made it go away, but playing just now I realized that turning down the PPIMV just made the output of the amp darker and made the fizziness less apparent.

I have a theory as to why the 100k grid resistance sounds better. I think that when the cathodyne runs out of headroom, the bootstrapping of the grid resistor stops, and the lower resistance loads down the driver more. So maybe more of the driver clipping gets thrown into the mix, and it sounds a bit rounder than when the driver sees a 1Meg load and keeps pushing the cathodyne past its boundaries.

I also tried moving the feedback wire to different OT taps. I settled on 8 ohms, which retained a bit more of the presence control’s treble boost and didn’t get as mushy as the 4 ohm tap. Overall this did not affect the quality of the onset of clipping as much as changing the PI grid resistance.

Disregard what I said earlier about the possibility of higher voltage or hidden ripple causing the poor sounding distortion. When I went back to the 1Meg grid resistor, it sounded just as bad as before.

Keep in mind, we’re relying on my 47 year old ears and perception for these conclusions…

 

[mickeydg5]

I always figured they increased Ra of the driver stage to 390k to bring the Q point closer to the proper voltage for the grid of the DC coupled cathodyne stage. Despite any extra gain the 390k might provide, it seems that most Orange guys prefer the sound of the self-biased arrangement. As I’ve read on the Orange Amp Field Guide website, Mike Soldano was a big proponent of modding late 70’s Oranges to the early 70’s self-biased cathodyne circuit.

The bias is mainly determined from the grid to cathode relationship.
The plate resistor is part of the tube's load. Higher values of resistance will promote higher gain level.

I do not know about the preference percentages of Orange users.

 

[mickeydg5]

I'm afraid I don't agree. A quick LTSpice model confirms for me that gain is very slightly lower with a small grid leak.

Probably, it's unusually high with the ECC83.

From a technical standpoint I feel the original design is the better design.

Show us your LTSpice views please.

 

[TheKman76]

I have a theory as to why the 100k grid resistance sounds better. I think that when the cathodyne runs out of headroom, the bootstrapping of the grid resistor stops, and the lower resistance loads down the driver more. So maybe more of the driver clipping gets thrown into the mix, and it sounds a bit rounder than when the driver sees a 1Meg load and keeps pushing the cathodyne past its boundaries.

The spice model looked to have some horribly sharp cathodyne clipping when driven. Not sure I follow your theory though, I was thinking along the lines of a grid current problem. The spice model didn't really give me an insight into *why* your 110K fix works and probably requires more thought on my part.

Regardless there are a number of ways to skin this particular three-legged cat. Was looking over some texts yesterday and a couple of them recommend big grid stoppers for cathodyne PI, 470K+. Being more-or-less unity gain there's no real capacitance to be concerned with. If this were my amp I'd probably lift some component legs and install a proper grid stopper to test.

Show us your LTSpice views please.

Sure thing. It will have to wait until this evening though (Australia GMT+10), I'm at work right now. Small signal sweeps, or transient?

 

[mickeydg5]

Yes, confident. Yes, directly to 115V tap.

Sure, but we’re talking an 8.3% change in filament voltage and a 23% change in HT. Also, I measured AC volts at the reservoir cap, and got 75VAC, which I guess is riding on top of the 411VDC. This seems excessive. I did rebias the amp before taking these measurements, and it did not improve performance.

This right here indicated a power supply rectifier or filter issue.
I thought this was talked about so did not pay more attention but I guess I was wrong.
I should have responded when I thought about it.

 

[FleshOnGear]

The spice model looked to have some horribly sharp cathodyne clipping when driven. Not sure I follow your theory though, I was thinking along the lines of a grid current problem. The spice model didn't really give me an insight into *why* your 110K fix works and probably requires more thought on my part.

Regardless there are a number of ways to skin this particular three-legged cat. Was looking over some texts yesterday and a couple of them recommend big grid stoppers for cathodyne PI, 470K+. Being more-or-less unity gain there's no real capacitance to be concerned with. If this were my amp I'd probably lift some component legs and install a proper grid stopper to test.

Yeah, I’m definitely feeling some hesitation to install a grid stopper. I’m concerned that simply lifting the cap and inserting a resistor will reduce gain too much, and I don’t like the idea of cutting a trace on this pcb.

The odd thing is that you’d think the frequency doubling distortion of the cathodyne would get worse as you increase gain, but I don’t hear that. The amp sounds badass when it’s wound up! My issue has been the edge of breakup.

I’m trying to think of how to describe my idea of what’s happening with the lower grid resistor. Think about the previous driver stage. It has a 100k DC load line. Now think of the AC load line, where the input impedance of the cathodyne is in parallel with the DC load line.

If the cathodyne grid resistor is 1M and we assume the bootstrap makes it look like 10M to the driver stage, what happens when the cathodyne clips? The bootstrapping goes away, and the driver sees 1M at the extremes. It goes from a 100k//10M AC load (99k) to a 100k//1M AC load (91k). That’s a difference of 8.2%.

If the grid resistor is 100k, it goes from 100k//1M (91k) AC load to 100k//100k (50k) at the extremes. That’s a difference of 45%. So, it could feasibly cause the driver to clip more readily when the cathodyne clips, thus softening the overall clipping.

Now, I understand that, when the cathodyne grid swings high, the cathode keeps going up as the grid draws current. But, it doesn’t follow the grid voltage as closely after that point, so the bootstrapping effect should be diminished. Also, there may be a softening effect at the negative excursion.

Anyway, it’s just a hypothesis. I could be way off.

 

[TheKman76]

Yeah, I’m definitely feeling some hesitation to install a grid stopper. I’m concerned that simply lifting the cap and inserting a resistor will reduce gain too much, and I don’t like the idea of cutting a trace on this pcb.

In my experience cathodyne clipping is yuck. Reducing input signal at the cathodyne can *only* be a good thing. Ideally this should cleanly reproduce what coming in, but the stage before it can produce far more swing that the cathodyne can handle.

I also hear what you're saying about AC loading the driver, just not sure I agree. Doesn't really matter providing you get the right result though.

For @mickeydg5 This was a very quick circuit half based on a completely different amp. Please ignore everything before V2A.
I've done an AC sweep and stepped the grid leak from 100K to 1M in 100K increments, green is the 100K line.
This is from Output1, but the same is true of Output2 and the signal at V2a anode.

 

[FleshOnGear]

Haha, I broke, @TheKman76. I looked at the guts of this amp, and all the rewiring and globs of solder from previous tech work, and I decided - screw it! I scratched out the trace to the grid of the cathodyne and wired in a resistor.

Unfortunately, I do NOT like the result. It smoothed out the edge of breakup very very slightly, but it robbed the full-on tone of punch and balls. It was too smooth and compressed. I tried a 100k, and then a 470k resistor there, but I’m going to remove the 470k and put in a wire.

I realize amps like this aren’t everyone’s cup of tea, but I really appreciate how it’s different from my Marshall. I don’t want to take away what makes it unique. I may or may not look into adjusting the balance of the cathodyne, but at this point I’m heavily leaning toward buttoning it up and just enjoying the amp.

 

[TheKman76]

Interesting! I agree man, make it yours and enjoy it. Did I forget to mention you could have lifted two components and make a wire bridge for this? I think I meant to, this would have saved breaking a trace.
Sorry about that, I have a flu coming on (winter here) and I may not be completely coherent.

You're leaving the 110K in there, right?

 

[FleshOnGear]

Interesting! I agree man, make it yours and enjoy it. Did I forget to mention you could have lifted two components and make a wire bridge for this? I think I meant to, this would have saved breaking a trace.
Sorry about that, I have a flu coming on (winter here) and I may not be completely coherent.

You're leaving the 110K in there, right?

Sorry you’re not feeling well. I always forget about the opposite seasons down there.

Yeah, I knew I could lift two components to do that, but I really don’t like leaving a bunch of ends floating like that. Oh, well. A previous tech already botched one of the pads for the 12AX7 socket and has a wire to the coupling cap, so this isn’t terribly out of place. Yes, the 110k is still in there.

Really, I’m enjoying the shit out of this amp. The distortion is primitive in a really pleasing way - like if a MkI Tone Bender or Maestro fuzz was transformed into a tube amp. And the tone controls can really shape the sound. You can go from an almost Vox-like brightness (with just a bit rougher texture) to thick bassy fuzz. And it has this punchy midrange bark that Marshall just can’t do. It has tons of gain, too.

My JMP is still my number one, though.

 

[mickeydg5]

In my experience cathodyne clipping is yuck. Reducing input signal at the cathodyne can *only* be a good thing. Ideally this should cleanly reproduce what coming in, but the stage before it can produce far more swing that the cathodyne can handle.

I also hear what you're saying about AC loading the driver, just not sure I agree. Doesn't really matter providing you get the right result though.

For @mickeydg5 This was a very quick circuit half based on a completely different amp. Please ignore everything before V2A.
I've done an AC sweep and stepped the grid leak from 100K to 1M in 100K increments, green is the 100K line.
This is from Output1, but the same is true of Output2 and the signal at V2a anode.

View attachment 133330

View attachment 133331

Nice. I was wondering about the gain differences.
I thought about downloading the program but did not know how long it would take to learn and setup.
So Output1 has less than a 1dB rise.
Can you please do the same for Output2 if it is easy enough for you?
And what is B1+ VDC?

 

[TheKman76]

This is a long way from a comprehensive model. In fact I've just ordered a text on modeling valve amps in LT Spice and making better valve models.

The small signal performance isn't a perfect indicator, but both outputs have identical traces and only 0.5dB lower than V2a anode. B1+ is around 320V.

LTSpice certainly requires some learning, but without all of the electronics equipment I had as a younger man this is what I have to work with. It's been extremely helpful in understanding circuit performance and testing ideas before modding stuff. I'd spend the time again without hesitation and intend to spend more time honing my results as it's *considerably* faster to model than build.

As an example, I purchased a Effectrode Blackbird valve preamp and modelled the whole unit trying to work out why I didn't like the sound much. It was supposed to be a Soldano sort of thing, but I didn't bond with it. Made some tweaks in the model, applied them to the Blackbird and viola! Love the sound now, very modded Marshall.

 

[TheKman76]

Really, I’m enjoying the shit out of this amp. The distortion is primitive in a really pleasing way - like if a MkI Tone Bender or Maestro fuzz was transformed into a tube amp. And the tone controls can really shape the sound. You can go from an almost Vox-like brightness (with just a bit rougher texture) to thick bassy fuzz. And it has this punchy midrange bark that Marshall just can’t do. It has tons of gain, too.

That's excellent man, I'm really pleased for you.

I get what you mean when you say 'primitive' too. There's something to be said for the really simple circuits with very few components and minimum gain stages, they do a thing which seems to get smoothed out to much in complex circuits. The DSL40CR for instance sounds great, but just doesn't do the raw, primitive thing well at all.