Issues with Printed Circuit Boards used in Guitar Amps

[loudthud]

I thought I would start a thread about mechanical issues facing guitar amps, in particular those built with Printed Circuit Boards.

First of all, some amps try to use a narrow PCB that collects leads from all the pots and input jacks across the front panel. Sometimes a few resistors and capacitors are included. I think the problem is created when enough mechanical stress is applied to the chassis to cause it to twist or deform ever so slightly and the stress is transferred to the solder joints between the pots and the PCB. Is there any way to avoid this? The first thing that comes to mind is a stronger chassis. The Fender tweed chassis is kind of strong just because it has a small left to right cross section. Bends in the sheet metal are close to the transformers which makes it more rigid than say a blackface era chassis. Some amps go a step further and build the whole preamp on the PCB. Is that any better or worse? Taking off all the knobs and pot's hardware is a PITA.

Another failure that occurs frequently is the leads on those snap-in filter capacitors. Are the PCBs not supported close enough to the caps to prevent or limit flexing of the PCB? Does squirting a lump of RTV silicon under the cap do any good? Does wave soldering just not put enough solder on the connections to make them mechanically strong? Is there a stronger solder?

I'm considering using some of those caps on a single sided PCB but installing eyelets on holes where the snap-in caps mount. Will that work? I may use 3/16 inch eyelets like MusicMan used on their PCBs.

On a related subject, I worked on a VOX amp recently that had a DSP board. The complaint was some sort of intermittent connection. Those DSP amps must have in the neighborhood of 1000 solder connections. I think a 100W Marshall tube amp has something less than 200. Isn't this a big reliability problem?

[Leo_Gnardo]

On a related subject, I worked on a VOX amp recently that had a DSP board. The complaint was some sort of intermittent connection. Those DSP amps must have in the neighborhood of 1000 solder connections. I think a 100W Marshall tube amp has something less than 200. Isn't this a big reliability problem?

All good thoughts & questions, and thanks for starting this discussion.

Lost in the mists of time, somewhere if I'm not mistook in WWII years, legend has it IBM undertook a study that brought to light what should be obvious: the more complex a system, the more likely it is to fail. MTBF Theory: Minimum Time Before Failure. Or was it Maximum? Anyway... Number of parts, number of connections, sure any one goes bad and you have a non functioning piece of equipment. Seems obvious enough, but they got it down to a "science" at least to the point where they could write some equations to describe the phenomenon. In the case of that VOX, definitely by having a super complex board, they're inviting disaster. From the manufacturer's side, they must expect a certain percentage of failures. Also they're expecting the amp owner to replace his/her gear every 5 years or so. Theoretically, used/failing gear goes to the dump. We folks on the service/repair side know those manufacturers are whistling in the graveyard - few amp owners look at the calendar and decide to swap out "old faithful" at the 5 year mark. And used amps continue to circulate in the second hand market seemingly forever, some with annoying intermittent problems that never do resolve.

Another look into the manufacturer's side - lots of companies build computers, right? Way complex boards in those, and most last 5 years or more until they wind up on the junkpile. We (insert manufacturer's name here) should be able to do that too, right? So it's OK to build an amp that's mostly a computer, stick some tubes in it for "authenticity," market that and be a howling success. Plenty of amps out there that match that description. But their owners seem to quickly lose their enthusiasm when they find the overall tone is naff, and reliability not so good. "Give me my old Twin back!" Or whatever amp they used to love and trust. I say it's best to keep the computer and amp separate. Get a Pod or whatever if you need that many tone choices, and a reliable, simple amp.

One last thought, a thumbs up to those manufacturers who try to make servicing their PC boards at least a little less of a PIA, by putting the traces and parts on the top side where we can get at them. Nobody's perfect, but Rivera comes to mind. I'm sure some other brands do the same. Still a PIA when you have to replace an op amp, but you can't have everything I guess. Good marks also to those who socket their op amps and IC's. Any time I have to unass a board to replace DIP packaged parts, I always put in a socket in case it needs to be done again. And in those cases, it never does... but we're prepared, right?

[alexradium]

usually the biggest poblems are on single sided boards and components not secured with glue/silicon,vibrations and shocks tend to fracture solderings.
If you make your own boards,i suggest make a separate pcb just for the critical power supply,where most likely you can have problems,both caps and diodes/resistors.
Otherwise make the project and send it to pro board makers and do it double sided so you have eyelets on every hole.
DSP boards also should be on a daughter board,so if you need a reflow of the processor you just release 4 clips and a connector,i've seen that on SS Marshalls,then the problem is just the connector that tends to move but you can secure it with glue.

[Chuck H]

If this is a design in the making and you plan to use PC pin mounted pots I've seen this work best with a narrow PCB dedicated to that purpose across the front panel and flying leads to the main board. The same would apply to jacks and switches. Of course that could end up being a bunch of smaller boards in the amp instead of one larger board. I'm not a repair guy, per se, but from what I've seen the problems start when these manual user, board mounted components have to share stresses with each other. Or worse yet, are actually the supports for a board that carries all circuits. Anyone here would instantly agree that amps with board mounted tube sockets sharing one PCB with board mounted pots that act as board supports is a failure waiting to happen. You didn't say this was a tube amp, but examples of what to avoid in this area abound and I've repaired a bunch of them. I like the idea of keeping the power supply board separate also. Though I haven't seen a lot of stress failures in shared board power supplies, usually more typical power supply failures that aren't stress related, but heat related. Most common I think is a lack of space between components that get hot, then derate, get hotter, derate more and then burn. So if it were me I would certainly try to provide some space/separation between components that get hot and also separation between components that get hot and those that shouldn't. And also elevate heat prone components from the board. Open air is often the difference between survival and a component derating into failure. Lot's of Mesa products have shown us this.

The most conscientious efforts used chassis mount pots, tube sockets, jacks and switches. Of course that's nearly impossible to achieve at any sort of competitive cost per unit. So I think doing the modular board thing is a good approach and the one I would use. One board for front panel controls, one for back panel controls, both narrow and and light weight, dedicate jack clusters greater than two, separate board for the power supply, perhaps even the power amp, and then flying leads to a central board where these all interface and, if applicable, preamp tubes are mounted. I know it seems like a lot more effort, but it's really not. Consider how fast construction could go just screwing on all the pots as one row or multiple jack clusters as one module and then plug it in or solder it to the central board. I think this sort of design also allows for better/cheaper serviceability, if that is a consideration, because modular board access and replacement would be a lot easier than lifting one entirely shared board just to replace something like an input jack, relay, etc.

[Mike Sulzer]

MTBF Theory: Minimum Time Before Failure. Or was it Maximum? Anyway...

Mean time between failure. Mean = average. It is a statistical thing. Who knows when it will fail, but on average you have so long.

[eschertron]

Regarding the 'computer-in-an-amp': are there many failures on the densely-populated boards due to mechanical stress? Or is the point simply that more parts and connections between parts equals more chance of failure?
I would think that PC manufacturers may not have a lot of data on how rugged these parts are, considering the normal PC doesn't get tossed into the back of a van 3 times a week. Laptops have a harder life, and I recall some "ruggedized" models advertised in past years. Not sure if they had anything on normal laptops besides a tougher case, but if data exist on these units - and laptops in general - on how they hold up it might benefit the design of amps guts as well. Just a thought.

[Gregg]

Most of the amps I built were PCB based (single and dual layer) with separate boards for PSU, preamp, pots, power tubes, FX loop etc so they weren't "one board amps". No mechanical stress issues so far. AFAIK such issues are not very common with the biggest manufacturers so maybe this is not a big problem after all.

[Leo_Gnardo]

Mean time between failure. Mean = average. It is a statistical thing. Who knows when it will fail, but on average you have so long.

Thanks for setting me straight on that Mike!

[g1]

This discussion could go a few ways depending on the intent of the build. Perhaps loudthud would want to clarify the production level, or maybe he just wants to hear all kinds of input.
Concerns for limited number builds will be much different than concerns for mass production. For mass production ease of assembly and cost of assembly will always trump serviceability issues. Reliability just becomes the failure rate % number.

[nevetslab]

Being a PCB layout designer over a good stretch of my career, having the end-user viewpoint and the maintenance/repair experience to be critical of all the stupid mistakes I encounter daily in servicing/maintaining amps built with PCB's, it has made me appreciate those few companies that will spend the $$ necessary for good quality raw PCB material whose foil adhesives are reliable, and even more when I see a lot of copper on the board, and NOT tiny annular rings for component pads, which are the default puppets in layout programs! If you're doing your job correctly, you're building your own pad stacks so there's enough copper bonding the fiberglass board, and will withstand numerous attacks from hacks with a soldering iron changing parts, and not lift off the PCB on the first attempt!! How often do we see such quality? I likewise get really frustrated with the dense-packed boards, such as found on GK gear.

As many in this thread have already stated, there's no excuse for using component terminals to mount a PCB, as though they're just as good as standoffs for mounting the board. The bean counters will always disagree, which is why we find PC mounted pots with NO support brackets and no mechanical mounts to take the stress off panel-mounted components, used to support a PCB and 'fly' it off the panel. Nope....no problem here....ship it!

The use of multiple boards IS a very sound idea...yes, more expensive, but, if properly done, FAR more reliable. I tend to be in favor of PCB wire-based connectors that use crimp terminals in their shells, over IDC-type interconnects. None of my thinking here is typical of being cost-competitive in the market. It's more based on seeing a product that is still working 20 to 40 years later, and is serviceable over that span.

My last PCB design efforts were for use in hydrophone array's, which once verified working, got potted and installed in the long array tubing. Fun project. I haven't gotten back to doing any OEM or limited production tube amp work, but no doubt will. One thought I've been having for a front panel PCB is using solder-lug pots, using discrete wires to the PCB, and mounted to a metal bracket attached to the PCB, so you don't have to unsolder the pot from the board to clean the pot or have to use bent-nozel tubing from the spray can in cleaning.

I do like double-sided boards, but.....hating blown-out plate-thru holes, I'd be very careful in the hole-sizes used for the components, so there's enough space to get the solder out the first time. I have nothing against wire jumpers on a single-sided board. Getting hot parts elevated, with enough space between them and adjacent parts is essential. Also essential in my opinion is leaving the copper on the board, and NOT flushing it down the drain. No reason to do that, apart from someone using auto-routing of a board layout program. If a board designer ISN'T also an experienced electronics engineer who knows low noise methods, minimal induction loop techniques and grounding, etc, he shouldn't be laying out PCB's.

[mozz]

Damm, who was that guy who did some Crate amps? I think it was a transistor guitar amp, what a nightmare. I forget his name but i guess he was a designer for Crate, psycho, they said you couldn't even talk to him. Putting resistors between ground sections. I think he thought ground loops and hum were preferred in designs. I remember cutting traces because it was a big ground loop.

[nevetslab]

All good thoughts & questions, and thanks for starting this discussion.

One last thought, a thumbs up to those manufacturers who try to make servicing their PC boards at least a little less of a PIA, by putting the traces and parts on the top side where we can get at them.

I'm glad you brought that comment up! How many times I've wished that when I have to lift up the main PCB in a Fender 65 re-issue Twin Reverb, then have to deal with doing my best to NOT break down the PCB adhesive in getting the bad parts off the board, and still loose the battle from their POS board quality! For sure the same thought in servicing HOT ROD series and their cousins.

You're right with regards to the IC in that approach. One can do those layouts as a double-sided PCB (higher cost board), and minimize what's on the bottom.

The first heat-sink mounted PCB's at BGW Systems...Model 250 & the 500D/750A were single-sided boards, with the foil face up. The gap between all the PCB and the heat sink surface was the nylon shoulder washers for the TO-3 power xsts. So, you had to be VERY careful when swapping out parts, so you didn't leave leads passing thru and hitting the grounded heat sink. It was before the day of our having silk screens on the boards so you could find the parts called out on the schematic, and we also didn't think to provide component placement guides. But, it was still just a generation from the garage operation where they began.

[R.G.]

The Workhorse line of amps contained everything I could think of at the time to make a PCB based tube amp quiet, easy to use and reparable. I put a lot of the considerations for PCBs inside guitar amps up in an article at geofex.com.

The considerations are many of the ones already listed. In general, guitar amps lead a hard life; high temperatures, mechanical abuse in frequent moves, large acoustic vibrations persisting for hours. They also have to work for decades, so need to be repaired. The features I came up with were, in no particular order:
> Make the PCB tough; I used 0.090" thick glass epoxy instead of 0.062". I'd have gone to 0.125"/3mm if I could have sourced it.
> Make the copper tough; I used two-ounce copper, twice as thick as most PCB copper.
> Make the board solidly mounted and well supported; I held the board into the chassis with sixteen screws all around the edges, no more than ... um... about 3" apart. I put two copper plated steel stiffeners down the length of the PCB to make it more rigid.
> Normally removing sixteen screws would be a service no-no, but I put a removable panel behind the PCB in the chassis so a service tech could access the solder side without removing the PCB from the chassis.
> Components were all through hole; SMD parts can pop off from board flex, and service techs hate SMD.
> Controls - pots, jacks, switches - were mounted on 2-3" long flying leads from the PCB. Replacing a control required only removing the back panel, which exposed the component side of the PCB, removing and replacing the pot/switch/etc. and moving the wires to the new part.
> The PCB was internally separated into ground domains by function, and those sections star grounded to the power supply.
> Components with any significant mass were glued to the PCB as well as soldered.

The idea was to sidestep many of the issues with PCB amps - having to remove every knob and jack-nut to get the %#$@^% thing out of the chassis, having to remove the chassis to work on the amp, having many PCBs communicating with each other on cables between boards, and having more layers of connectors for those cables.

Keep it simple, short, and direct. Make the wear parts - tubes, pots, jacks, switches - easy to get to and easy to replace.

[Mike Sulzer]

After thinking about how to do all those things really well, some people might justifiably conclude that it is better not to use a PCB.

The Workhorse line of amps contained everything I could think of at the time to make a PCB based tube amp quiet, easy to use and reparable. I put a lot of the considerations for PCBs inside guitar amps up in an article at geofex.com.

The considerations are many of the ones already listed. In general, guitar amps lead a hard life; high temperatures, mechanical abuse in frequent moves, large acoustic vibrations persisting for hours. They also have to work for decades, so need to be repaired. The features I came up with were, in no particular order:
> Make the PCB tough; I used 0.090" thick glass epoxy instead of 0.062". I'd have gone to 0.125"/3mm if I could have sourced it.
> Make the copper tough; I used two-ounce copper, twice as thick as most PCB copper.
> Make the board solidly mounted and well supported; I held the board into the chassis with sixteen screws all around the edges, no more than ... um... about 3" apart. I put two copper plated steel stiffeners down the length of the PCB to make it more rigid.
> Normally removing sixteen screws would be a service no-no, but I put a removable panel behind the PCB in the chassis so a service tech could access the solder side without removing the PCB from the chassis.
> Components were all through hole; SMD parts can pop off from board flex, and service techs hate SMD.
> Controls - pots, jacks, switches - were mounted on 2-3" long flying leads from the PCB. Replacing a control required only removing the back panel, which exposed the component side of the PCB, removing and replacing the pot/switch/etc. and moving the wires to the new part.
> The PCB was internally separated into ground domains by function, and those sections star grounded to the power supply.
> Components with any significant mass were glued to the PCB as well as soldered.

The idea was to sidestep many of the issues with PCB amps - having to remove every knob and jack-nut to get the %#$@^% thing out of the chassis, having to remove the chassis to work on the amp, having many PCBs communicating with each other on cables between boards, and having more layers of connectors for those cables.

Keep it simple, short, and direct. Make the wear parts - tubes, pots, jacks, switches - easy to get to and easy to replace.

[R.G.]

After thinking about how to do all those things really well, some people might justifiably conclude that it is better not to use a PCB.

They might. No one ever said that good PCB design is easy, especially when just laying down components and routing copper traces is the lowest, most trivial part of what it really takes. In doing a good PCB, you're considering the mechanical layout, locations and organizations of controls and such, signal flow, cabling needs or not-needs. It's much more system design, not just PCB design.

On the other hand, these concerns are not limited to PCBs. Hand wiring has its own set of things that have to be considered to do it well. You still have to worry about what signal is in which conductor, and how to get them all from and to. There's a hierarchy of needs for making good hand-wired joints, wire routing, and so on. I suspect that hand wiring simply seems easier because there are so many prior examples, and for the idea that you can always rip it out and start over.