Ampeg B15 rebuild

[Embenny]

Man, I appreciate your wiring jobs so much. They're works of art. You put a really high level of care into your gear in places that nobody else's eyes will ever see (except us, since you photograph them).

[øøøøøøø]

Thanks, yeah it helps to be a little fanatical about keeping it neat... a quiet amp that doesn't oscillate or have any other weird problems is down in part to how the wires are physically laid out. It's not just for vanity (but if you're being careful anyway, might as well also make it look good)

[øøøøøøø]

Time to actually (re)build the amp. Starting with the power supply, I connected the transformer.

In many ways old Ampegs are engineered so differently from old Fenders or Marshalls. I really like how this amp handles standby--it simply lifts the center tap of the B+ winding from ground.

I actually had to desolder the standby switch after this picture was taken because I'd forgotten that those wires needed to run through the provided channel in the wooden standoff under the tag board. Better to do the rework than to do what the last person did and simply grab the end of the wooden spacer with something and break it off...

I added an isolating terminal on one of the PT mounting bolts to give me a little more flexibility with grounds. I'm trying to keep the amp mostly stock-looking (i.e. not adding a bunch of bus wire) and avoid further non-reversible alterations, but i'm trying to create a scheme that will function substantially like a ground bus in practice. We'll see if I succeed.

Multi-section filter cap is isolated from chassis (as it was from factory). It connects to this isolating terminal, which will hold the ground points from the filament and B+ center taps, the reservoir capacitor ("first filter cap"), bias supply ground, and the cathodes of the output tubes. Nothing is connected to chassis ground yet--I want total control of that.

Earlier stage grounds will run back to the multi-section cap (not ideal, but chassis layout boxes me in) and then ground will connect to chassis at one point--the input jacks. I chose this point to avoid enlarging the holes in the (substantially-original) faceplate, which would've been required to use isolating shoulder washers for the input jacks.

We'll see how it works--it may do what I want and make for a very quiet amp, or it may do the opposite of what I want. I can always change it later.




Speaking of chassis layout boxing me into non-preferred practice--the output transformer sits right under the sensitive input stages, and the wires must run right through. Having read that the later PCB versions of the B15 have a shield through which the OT wires run, I decided it's best not to take any chances--I twisted the OT primary wires together, bundled them with the OT secondary wires, and surrounded the whole thing in metal braided shield. I think covered this whole cable assembly in heat shrink tubing to which I soldered a ground wire (and also the wire for the OT's internal shield).

Along with the shielded grid-circuit wires in the input stages, this should buy me a little insurance against parasitic oscillations.

After connecting that all up I verified that I have no continuity to chassis anywhere yet. That's good; it'll give me a lot of control (and also flexibility) over how the ground connections get made. This will exclusively be a recording amp, so might as well make it as quiet as possible.



Power supply filter capacitors are connected (except for the enormous 800v reservoir cap, which I've yet to mount).

OT primary leads connected to pin 3 of each 6L6GC. I phased them the way the amp was when it came to me (working): Brown on V4; blue on V5. Strangely, this is the opposite of what I see in most B15 chassis pictures, so I'll have to watch out for that. If the negative feedback becomes a positive feedback loop and starts oscillating/squealing, I'll know why--then I can either swap the primary wires or return the 10k negative feedback resistor to the opposite side of the phase inverter.

The preamp section is wired up now as well--shielded wires to the front panel jacks and controls (i.e. the highly-sensitive grid circuits) as from factory. I've done cathode wires in purple and plate wires in red. Trying to avoid long parallel runs of substantial length, and separate them a bit where it's unavoidable (inverse-square law is our friend again). But this may be slightly academic: the big oscillation threats are the grid circuits, and they're shielded.

[øøøøøøø]

The current state of things--getting close! Just have to go through with a schematic and methodically check everything.

I've made a few changes to the layout since last time--I'd located the output tubes' 1k grid stoppers on the eyelet board, which isn't the best place for them... they should be directly on the tube sockets with as little lead length as is practical, so I made that change. I ran some wires to the board; they'll be mostly-covered by the last two capacitors when they arrive.

Two deviations from stock:

1) I've added 1 ohm, 1% resistors between output tube cathodes and ground (using the n/c pin 1 on the tube socket as a convenient tie point). This will allow me to conveniently measure output tube bias without getting the calculator out, and without placing meter probes in high-voltage areas. If you've ever used a commercial plug-in biasing device, they use this method, which is as accurate as your resistors are precise.

2) I semi-reluctantly abandoned the slick stock speaker impedance-switching configuration in favor of something more conventional.

To reach that second decision I weighed out two competing factors: how comfortable I am sending my entire signal to speaker through a mechanical connection that can corrode, versus how often I'll be likely to connect a second speaker cab. I reluctantly decided it wasn't worth the tradeoff. An external speaker can still be connected, it'll just show the OT the parallel load rather than a series load (like a Fender amp).

The Ampeg solution (to place the speakers in series and use the leaf spring on the isolated Ext. Speaker jack as a "normal" for the lower-impedance tap) is some great engineering. But I'll essentially never connect a second cabinet, so the Fender-style solution (which has the benefit that speaker connections are soldered and can't become intermittent or gain series resistance due to corrosion) seems like the better trade-off this time.




Here's an annotated picture of how I've run my ground/earth connections. Hope it's as quiet as I think it'll be! I'm not an "amp tech" (much less a design engineer), so I'll see whether theory/received wisdom deviates from practice.

1960s Ampeg had some motives I don't have--manufacturing expediency, cost, etc. This boxed me in to a few trade-offs, but I still think this setup should be a bit more optimized than the stock configuration.

My goal was to lay ground connections out in a bus-style configuration (terminating in one location near the input jacks), but without modifying the amp to use a literal bus bar.

The biggest challenge was the can capacitor, which serves as the filter capacitor for the screen supply, the phase inverter, and the preamp. That means all of these capacitors have a shared ground, and there's nothing I can do about it (short of modifying the amp for discrete capacitors).

The good news is that the can cap is isolated from chassis, and that the reservoir cap ("first filter cap") is separate (it's the large black "firecracker" ziptied to the plywood chassis side on the left). Both points are very important, and give me hope that the layout can be really quiet.

The reservoir cap should form a tight physical location with the power transformer and rectifier tube, and all grounds should be connected together for a short return path between those elements. This was done at point "D" on the illustration... an isolating terminal strip that does not connect to chassis.

A ground wire from the hum balance pot (filament supply ground) also terminates here, as does a wire from the bias supply ground toward the left side of the eyelet board. The ground wire from the hum balance pot is routed directly over the longest run in the heater supply run to give a touch more shielding.

Point "C" is the can capacitor... shared screen supply, phase inverter and preamp stage grounds. Luckily, the ripple currents in the screen supply are far lower than in the reservoir cap/plate supply, so this probably won't be too catastrophic. This connects to point "D" with a single 18ga wire. We still have no continuity to chassis yet!

Point "B" is the phase inverter's ground. It holds that tube's cathode connections as well as the output transformer and speaker ground (though the protection mechanism contained in the 4-pin XLR speaker jack). Since the amp uses negative feedback that returns to this stage from the output transformer secondary, their grounds should be together at this point. This connects via a long 18ga wire to point "C." Still no chassis connection (verified at each step with meter).

Next we take a single wire to join point "B" to point "A," which holds the preamp tubes' cathode grounds. Then with one more direct wire connection we join these with the input jacks, which ground themselves at the chassis mechanically via locking washers... the one and only chassis ground connection in the entire amp.

The input jacks are securely tightened with lockwashers on the rear getting a good "bite," but I also connected all three input jack grounds together to give some extra insurance... for the amp to lose its ground reference, all three jacks would have to simultaneously be loose. This is not impossible, and on a commercial product it wouldn't be the way to go. Far better to make a solid chassis connection (preferably soldered), isolate the input jacks, and connect them to chassis ground via a .01µF capacitor (to kill interference).

But rather than do the irreversible mod of enlarging the input jack holes for shoulder washers, I settled on this compromise. With the input jacks all three cinched down really tightly, the connection (with redundancy) ought to be plenty reliable for me.



Also from fliptops, I got a new branded Lucite panel and some gasket material for the tube cage. The old stuff had totally perished... rock hard, and it was also painted (looks like someone painted it at the same time they painted the cab!)

[countertext]

I love the detailed info in this thread. I’ve got a Gretsch 6151 that needs to be rebuilt, and even though I don’t understand 90% of what I’m reading here, it’s so much smaller and simpler (I think) than this B15 that I think I might be able to do it.

[sessylU]

What's that weird honeycomb pattern? Is that just a fluke of the angle you've taken the photo?

Can't work out if that would drive me mad or cause me further infatuation.

[øøøøøøø]

Totally just a photo fluke!

But it does look kinda crazy

[øøøøøøø]

Looks like Zuckerberg has figured out I'm linking to photos and shut it down; too bad.

Anyway, I got the last few capacitors and a nice pair of red-base TAD 6L6GCs and and got the amp ready to fire up.

I pulled the Eminence Delta 15a speaker out of the "original" cabinet (which I've become halfway-convinced isn't original at all, but rather a homebrew made to fit the original lid and chassis tray) and alligator-clipped it to the speaker output. I also hooked up a scope across the output and used another clip lead to make the PI/OT ground that's usually made when the 4-conductor speaker cable is connected to the Portaflex cabinet (a protection mechanism to prevent the amp from smoking the OT if powered up with no speaker connected).

First item on the list was to adjust the hum balance pot. It must've been super corroded/dirty inside because it was momentarily disconnecting itself with every turn, and exercising it alone wasn't enough. A shot of DeOxit D5 followed up with a few drops of F100L and it was good as new.

Adjusting hum balance was interesting--on the scope you can see a sharp spike every cycle which I can only assume is something related to the utility power. This obviously generates a lot of high harmonics.

Adjusting the hum balance downward, you can see this spike reduce in amplitude at a rate faster than the sinusoid portion of the 60 Hz signal, until it disappears completely and the sinusoid starts to increase in amplitude. The quietest sounding position (right now) is the point where the high-harmonic "spike" is all the way gone, even though the 60Hz fundamental component is actually a little bit higher in amplitude.

Will this change once the speaker is in a cabinet and can better-reproduce the 60Hz fundamental? Possibly. Here's a video of turning the pot:

https://www.youtube.com/shorts/Lgo6UIzX0AQ

Then I set about going through the amp controls to make sure everything worked as expected, and there were a few little issues... namely a bit of RFI (radio station pickup, to be more precise) and other odd behavior at certain knob positions... moments of loud buzz, edge-of-oscillation type of behaviors, etc.

My first troubleshooting step was to disconnect one end of the 10k negative feedback resistor to rule out output transformer phasing issues... no change, so this was not the cause of the issue.

Then I remembered something curious about (what I thought was the) original shielded wire when I had to re-do one end: the "shield" didn't seem to surround the internal conductor, but was more like a second bare conductor that ran alongside. So I grabbed a chopstick and started pushing "shielded" wire around and... I could get all of the issues to intermittently appear and disappear.

So I ordered 20 feet of proper 24 AWG single-conductor shielded wire and my next step will be to re-do all the shielded runs in the amp.

Not too big of a setback since I've probably still got another week or two to wait on the cab, anyway.

[OffYourFace]

I want to hear it when you're done!

[øøøøøøø]

I do too!

[hulakatt]

Any progress? Either on the amp or rehosting the pics?

[øøøøøøø]

Interesting progress on the amp! It's fully-working now, but still waiting on the cab.

When I have a second I'll walk through an interesting process I had to go through to get it to behave, and some lessons learned in the process.

No progress on (or much thought given to) rehosting the pics yet.

Back to editing now...

[hulakatt]

I've been happy with Flickr since the great photobucket implosion

[Dr Tony Balls]

I've been happy with Flickr since the great photobucket implosion

Same. I just pay for a pro account or whatever its called and it works great.

[øøøøøøø]

Okay so here's the journey this amp sent me on. It actually wasn't that bad, but it was interesting.

I decided to elevate the heaters off the bias supply instead of grounding them, which is something I found in a later schematic revision. This made things just a little bit quieter; worth it. While I was at it I also replaced the hum balance pot with a nice old surplus 5W wirewound type with a locking bushing. That way I can get the heaters quiet and then lock it down so the trimmer doesn't inadvertently get bumped.

As described above I decided to reconfigure the grounds into a pseudo-bus ground, as best as I could with the multi-section cap serving as the PI and preamp power supply nodes.

On test I had some serious RFI issues (I could tune in radio stations and had weird clicks and buzzes at certain knob positions that I could change by pushing the "shielded" wire around).

I put "shielded" in quotes because this wire didn't appear to have the bare conductor surrounding the insulated conductor, but rather running alongside it. So I bought some proper shielded wire... PTFE insulated inside and out with the internal conductor fully-enclosed in braided shield.

And.... I still had the exact same RFI issues!

I'm a fucking guitar player, not an electronics engineer... so I had to do some reading to understand that at very high frequencies (like radio frequencies) a single-point ground doesn't behave the same way as it does with audio and line frequencies. When the frequencies are that high, inductance and resistance of a length of wire becomes significant, and the very-high-frequency stuff will try and find a path to ground through stray capacitances... this results in RFI pickup.

Since the input jacks are the location of my single chassis ground, they were definitely not the source of the RFI issue. The issue was the second stage... the stage fed by the volume control. This much was clear. The volume controls actually had a substantial length of wire before they connected back to the chassis, and the resulting impedance to chassis ground was apparently enough to turn it into a radio receiver.

My solution was to slip a solder lug over each volume pot bushing and solder a .01µF cap from the pot's ground to chassis ground very close to the pot itself.

Bear in mind that this isn't a mod to the circuit of the amp at all... it's just a little ceramic cap that sits between two points that are both at ground potential.

As far as the amp is concerned, It's literally no different than just soldering both leads of a cap directly to the chassis... it's just done in such a way that gives the RFI a sensible place to land instead of trying to infect the signal. And the cap is small enough that it won't pass 60Hz or 120Hz ground currents, or (presumably) audio frequencies. So the "single-point" ground remains in effect at all frequencies of interest. Really pretty interesting!

Now the amp works perfectly (V1 is slightly microphonic, but that's easily fixed). It's very quiet for a B15... channel 2 is slightly quieter; I'll have to investigate that (it might be that V1 acting up).