There is PCB and then there is PCB

[defaced]

Never mind. I had an older browser window and RG already answered the question.

[GainFreak]

It's a mechanical issue. The thermal coefficient of expansion of copper and glass-epoxy are different. The longer a parallel run of copper and glass-epoxy, the more stress temperature changes put on the bond between the PCB stock and the copper trace. PCBs with unequal top and bottom side coverage tend to bend with large temperature changes, just like, and for the same reasons as, bimetallic thermometers and thermostats. Bending the trace ever so often lets the trace have some bending compliance that straight runs don't have. It doesn't have to be much. Just some.

Wider traces have less problem with the copper delaminating, as there is more surface area of adhesive to the PCB in a given length of run. Thicker copper makes the copper stronger and needs a wider trace to avoid overstressing the adhesive.

It's not an issue in hand wiring or ribbon cables.

A long, thin trace doesn't automatically fail, or even mostly fail. But if you have the space, a couple of wiggles will ensure that it almost never does.

Just one of those things you pick up over time.

Thanks for this detailed answer. I guess this will apply mostly if we're talking traces for tubes' filaments?

[R.G.]

Thanks for this detailed answer. I guess this will apply mostly if we're talking traces for tubes' filaments?

There's a lot of "well, it depends" on this one. Self heating of a trace, which is what I think you were thinking of, would be one issue. But a well designed trace does not significantly self heat.

Just a long trace presents the problem if the board can flex. On the amp boards I did, I star grounded the PCB; every circuit section had its own local ground domain, and that ground domain had its own ground trace back to a star ground at the negative of the first filter cap. The ground wires didn't carry significant current, but could have been prone to cracking or tearing loose from solder connections if nothing was done about flexure stress-strain. In some places there was room to wiggle the trace every couple of inches, and in some places not. The PCB was in a head with the tubes providing their own private oven, and measured air temps went up to 58C. I solved this one by putting two steel ribs down the long dimension of the PCB. No amount of heating would flex the PCB, so the trace copper didn't get a chance to crack with flexure.

Copper work hardens. Flex a PCB a little and over time the copper hardens and eventually will crack. Solder work hardens worse than copper. For reliable boards you have to figure out how to not flex that stuff.

When I was doing research for the amp, I interviewed some grizzled old amp techs and asked them to show me PCB amps they didn't like.

One guy pulled out a control board from a M*****ll. It was 3" wide and 19" long, had 20 PCB mounted controls/jacks on it which were soldered directly to the PCB, and when held in the middle and moved up and down, flexed like birds' wings flapping. It was almost an example of what not to do: thin PCB material to make flexing easy, controls hard-soldered to the PCB to transfer operator movements to stresses on soldered joints, large number of controls to remove before any repair action was possible, special parts needed for controls, long skinny PCB not supported mechanically.

Our amps were intentionally designed to be the opposite. Thick PCB, board-mounted stiffening ribs, thick copper, all controls on flying wires, access to all components without removing the PCB from the box, and the ability to replace pots and jacks with standard parts without even removing the chassis from the box.

It's very possible to design reliable and easy to service tube amps. If you want to.

[Steve Conner]

It's very possible to design reliable and easy to service tube amps. If you want to.

Sure. Selling enough of them at enough of a profit to stay in business is the hard part.

According to your earlier treatise on economics, we can deduce that the market actually wants the Bugera, Marshall Valvestate and Line6 crap, because that's what sells the most. All the corners that they cut in design, they have to cut to stay in business.

Of course, I don't really believe this and suspect you of sleeping with Milton Friedman.

[Chuck H]

It's very possible to design reliable and easy to service tube amps. If you want to.

It helps the design process A LOT if you've serviced and designed a few already though. All the good/bad examples you set forth aren't recognized by the consumer, as evidenced by the recent rash of new members here with less than 10 posts that want free service on their Behringer, Crate, BlackHeart, whatever. So it seems your outline is a good one for novice and experienced designers alike. For small operations it's not cost efficient to design amps that will break. But for big MFG's it seems to be a very different story.

[jrfrond]

First of all, I am not a big fan of PCB's for tube amps. Please don't give me the positive arguments for them. I've been building, repairing and modding guitar amps for over 30 years, and I have my own reasons for liking turret/terminal boards and PTP hard wiring.

One of the reasons a lot of PCB-based amps suck is that, often, the reason they exist in that domain was as a result of cost-cutting. The cost-cutting measures don't stop at PCB design, but goes all the way down to component level, and man, I've seen some of THE CHEAPEST shit components in amps nowadays, even flagship stuff from major manufacturers! They don't SKU, for instance, separate 100K plate resistors for the handwired amps. They use the same garbage all the way around, and THAT folks is what the problem is nowadays. As much as PCB's for tube amps aren't my thing, I won't blame the demise of "good" guitar amps on the PCB's alone, unless the boards are just of plain piss-poor design. It's the component quality that REALLY sucks. The REAL difference behind a manufacturer's reissue and a boutique copy lies mostly in component quality. Blame it on the bean-counters, not on the design engineers.

Unfortunately, this issue doesn't only affect guitar amps, but literally EVERYTHING we work on.

[Steve Conner]

One of the reasons a lot of PCB-based amps suck is that, often, the reason they exist in that domain was as a result of cost-cutting. The cost-cutting measures don't stop at PCB design

+1. Like I said earlier, the problem with PCB amps isn't necessarily the PCB.

[GregS]

Copper work hardens. Flex a PCB a little and over time the copper hardens and eventually will crack. Solder work hardens worse than copper. For reliable boards you have to figure out how to not flex that stuff.

Other ways a PCB gets flexed besides regular use is during repair. When I re-soldered the power resistors in my Fender Blues Deluxe to fix the random channel switching problem I had to turn the board around. The board was stressed doing this due to a large number of flying wires attached to it. Afterwards the channel switching problem was fixed but a number of other intermittent/bad sound problems appeared - probably from cracking copper traces and solder connections. I decided it wasn't worth it to spend any more of my time on that amp and just use it as a test speaker cabinet for my builds now. I see repairs as a problem for cheap circuit board amps in more ways than one.

Our amps were intentionally designed to be the opposite. Thick PCB, board-mounted stiffening ribs, thick copper, all controls on flying wires, access to all components without removing the PCB from the box, and the ability to replace pots and jacks with standard parts without even removing the chassis from the box.

So you can remove or replace parts without having to remove the board and turn it around?

The REAL difference behind a manufacturer's reissue and a boutique copy lies mostly in component quality. Blame it on the bean-counters, not on the design engineers.

Not disagreeing with anything you're saying. Just to add to it this is the reason one can get a tube amp for $500 today which would probably be the equivalent of $200 or less 30 years ago. Back in those times tube amps were not nearly as affordable. However you get what you pay for and one of these modern cheap amps will not last as long and be more difficult to fix when there is a problem.

[Gregg]

Just a long trace presents the problem if the board can flex.

I have few PCB designs of my own some of them 90x480mm big. If you have 2mm thick 70um copper double sided PCB fixed with metal standoffs in 7-10 points to the chassis it doesn't look to bend anytime soon.
What I'm using for controls is a separate narrow PCB, pots mounted and fixed to the front panel which excludes transferring stress on the solder joints, also this way you can easily remove the main board or the pots board if necessary.

First of all, I am not a big fan of PCB's for tube amps. Please don't give me the positive arguments for them. I've been building, repairing and modding guitar amps for over 30 years, and I have my own reasons for liking turret/terminal boards and PTP hard wiring.

It depends. If you are as I call them (without any bad feelings) 3 or 5 knob (boutiqe) amp builder or manufacturer you can forget about PCBs. But if you design a 3 or 4 channel amp along with plenty of extra things on it you simply can't do it without a PCB.

[kg]

for those interested in doing their own PCB work, keep in mind that it WILL sound different to a PTP wired design. blame that pesky stray L and C.

[big_teee]

On the Topic of PCBs.
Tube Depot make 2 tweed amp Kits that are more or less Point to point except for a PCB they use instead of a turret, or Eyelet board.
I'm mainly talking about their 5e3 Deluxe tweed kit.
My question is what will be the long term value of the amp with the PCB vs. a Amp made with a turret or Eyelet board?
On hand built stuff there's a lot of emphasis put on point to point.
Point to point is usually considered by many to be anything not on a PCB.
Here's a picture of the board in question.
Comments, and opinions welcome.
Thanks in advance.
Terry

[Gregg]

for those interested in doing their own PCB work, keep in mind that it WILL sound different to a PTP wired design. blame that pesky stray L and C.

Well I'm definitely not building PTP a 3 channel amp with tons of switching and MIDI just to compare it how it sounds to a PCB design. At least I don't know of such amps being built PTP.

[kg]

Well I'm definitely not building PTP a 3 channel amp with tons of switching and MIDI just to compare it how it sounds to a PCB design. At least I don't know of such amps being built PTP.

all i'm saying is that IF you breadboard a design and tweak topology and component values until things are JUST RIGHT, THEN try to simply slap the thing on a PCB, you will be disappointed.

nobody would say lead dress doesn't matter in an amp (well, nobody intelligent at least). this is the same concept, except there's no going back and changing it once it's printed.

edit: of course the reverse is also true: trying to copy a PCB design as PTP will result in a different sounding amp. how different? don't know till you try...

[Gregg]

You know I'm not exactly a theory guy although you should know some basics so I think simple. If my design doesn't hum, squeal and/or produce any abnormal/unexpected sounds/noise it's OK for me. If it works for 5-10 years or more fault free even better. Can it be done better? Most probably yes.

[R.G.]

nobody would say lead dress doesn't matter in an amp (well, nobody intelligent at least). this is the same concept, except there's no going back and changing it once it's printed.

Likewise - there's no making another one just like the hypertweaked one that's just right with hand methods. Almost like you have to choose your poison.

edit: of course the reverse is also true: trying to copy a PCB design as PTP will result in a different sounding amp. how different? don't know till you try...

I'd put it differently. PCB's strong suit is repeatability. So I'd say if you ever did get a PCB design just the way you liked it, you could do it again. The tweaking works both ways; PTP (which is NOT the same thing as hand wired, tagboard, turret, or eyelet: those are different still) has a wider variation because of those pesky Ls and Cs between the wires that all get kind of bunched together, more or less. The position of every wire matters, more or less. Good PCB design arguably has the potential to get the Ls and Cs of the strays within the range of those found in hand wired formats. Once you get them there, they stay, within the ability of PCB etching to place a trace within +/-0.001" and manage its width and pathway within about the same amount. Hard to do that with hand wiring.

Point to point is usually considered by many to be anything not on a PCB.

Yeah. And everyone who had to solder even one wire inside the box says it's "hand wired".

So you can remove or replace parts without having to remove the board and turn it around?

Yes. Six screws drops off the backing plate and you can access every solder joint without removing the PCB from the chassis.

Sorry - I could not figure out how to do this without removing the chassis from the amp, but it's only for PCB mounted Rs and Cs that this is needed. All the controls, switches, jacks, tube sockets, etc. are accessible and replaceable with the chassis in the box. Just drop off the back panel with the chassis still in the box. You don't even have to pull off the knobs except for the one you're replacing.

Blame it on the bean-counters, not on the design engineers.

I refer to this tendency as "MBA Disease".

for those interested in doing their own PCB work, keep in mind that it WILL sound different to a PTP wired design. blame that pesky stray L and C.

Tee hee.

For those interested in doing their own PTP work (or tagboard, or eyelet or turret board!), keep in mind that the position of every wire and component will be different on each "copy", so every single one will sound sound slightly different from each other, as well as different from a PCB version. Blame that pesky stray L and C.
Worse yet, what do you do when the owner complains that it doesn't sound the same, and all he did was move some wires around?

A PCB version *can* be made to fall within the range of variation of the handbuilt ones, and the PCB versions will vary much less from one another than the hand wired versions. For this we can blame the pesky consistency of the stray Ls and Cs on that PCB.

Not that you can't do a PCB layout wrong, or that it may be difficult to do correctly, but once you get it right, each one comes out nearly identically => if you do it right <=.