Spud: what, again!?!

[squiffiness]

Originally Posted by Fenris

I have had zero problems with the service I linked to. I'm attaching a picture of two boards I had done a year ago (actually, one 10cm x 10cm panelized board that I cut out myself). The fit and finish is professional and they solder just fine. I just ordered another 10 of a 5 x 5 cm board, I'll let you know if the quality has been maintained.

Hey Fenris,

Thanks for the information for the thread. Was your cost for the boards you attached $9.90 for two, or 10 for $70?!?

What about tooling setup, drill holes, hot air leveling, etc. to get a finished board?

The product you have looks professional. How was their turnaround time and shipping costs?

[Fenris]

The cost is $9.90 for ten 5 x 5 cm boards, it's $24.90 for ten 10 x 10cm boards, and $69.90 for ten 10 x 15 cm boards, as well as other sizes (there's a drop-down to select size, and price varies according to size). Standard color is green, other colors available for a $10 charge. Hot air solder leveling is standard, ROHS and/or gold plating for an additional charge. Electrical test of 50% of the boards is standard, 100% for an additional charge. Honestly, the basic options are fine unless you really want a super cosmetically pretty board (black PCB with gold plating...mmmm). Shipping is in addition to the price, but for ten 5 x 5cm boards, shipping is $4.10 for airmail. Time frame is 10-30 days, depending. Faster shipping is available, but I'm not going to spend $40 on shipping $10 of boards.

With 9 pin Novals or Novars, I'm sure a 10cm x 10cm (slightly less than 4" square) board could be designed. I've designed a 6L6 driver board using two 6FD7 tubes and it's smaller than 10cm square. Granted, there's no power supply onboard, but I like using the big square diode bridges offboard and regulator PCBs. A separate power supply board would be cheaper than a single big PCB.

[vinylkid58]

Originally Posted by tubelab.com

Watching that stupid reality show has taught me how to survive without the heart attack that was coming my way about 12 years ago.

Maybe not such a stupid show after all.

jeff

[squiffiness]

Originally Posted by Fenris

The cost is $9.90 for ten 5 x 5 cm boards, it's $24.90 for ten 10 x 10cm boards, and $69.90 for ten 10 x 15 cm boards...Hot air solder leveling is standard, ROHS and/or gold plating for an additional charge. Electrical test of 50% of the boards is standard...the basic options are fine...Shipping is in addition to the price...Time frame is 10-30 days, depending...

That's a pretty amazing price, all things considered. In what units (mils or mm) did you design your PCB's?

What software did you use? Looks like the service likes a Cadsoft Eagle Design Rules check. Since I am new to all this, I'm trying FreePCB. It has a design rules check, but I am too novice to know whether it's check is consistent with the Eagle check AND whether the Gerber files it will produce are a format compatible with your fab service.

Originally Posted by Fenris

With 9 pin Novals or Novars, I'm sure a 10cm x 10cm (slightly less than 4" square) board could be designed...no power supply onboard...A separate power supply board would be cheaper than a single big PCB.

If George "Tubelab" is finalizing a design, I suppose that's up to him, but the theory of an optimized "Tubelab SE Power Supply" board with some extra board edge connectors to carry over B+, filament supplies, pilot light feed, etc. would be nice features, but not "must haves" for me.

Ultimately, at $7 a board for a batch of ten 10cm by 15 cm boards, which the SpudSE proto-board is (roughly), I am hoping that we can coax George to pull a potatoe (thanks Dan Quayle) out of his hat, after pulling his Guitar Amp board out for his audience.

[tubelab.com]

Tubelab, if you want a board as an example, let me know, I'll drop one in the mail.

Thanks for the offer, but it is not necessary. I have 3 partially completed board designs on my computer right now. The Spud SE would be the 4th. I will try this PCB service with whichever board I finish first, just to see how good they are. 3 of them are my typical tube amp designs which are very easy to make (large holes and pads, fat traces and wide spacing). The 4th is a fairly dense 4 layer SMD board mixing RF, digital, and analog circuitry. Seeed doesn't do 4 layer, or it would be a good test of process capability.

I think you should be building boutique guitar amps for people as your sideline/retirement.

I did that for a while about 30 years ago. It takes a while to build up a reputation and clientel, which I have long since lost, since I have been away from that world for a long time. It also takes a bit of work and time to find out just what the perspective client wants. How do you deal with a guy who walks in and says "I have this cool sound in my head, can you build it for me?" If you live in a big city with several large "hands on" music stores, you meet him there and crank up several amps and say "show me". Here (where I will likely retire) the closest big music store is a 1.5 hour drive!

There will likely be some guitar amp kits though.

The 2008 econ-implosion decimated my 2005-2010 company (580 laid off in Toronto over 16 months).

We had 353 people laid off in a single day. In 2000 there were 140,000 worldwide employees. Since then there have been layoffs, spinoffs, selloffs, and the ultimate splitting up of the company. Once the cell phone division acquisition by Google is complete, there will be 23,000 worldwide employees.

There are 3 major corporations designing cell phones in south Florida. Motorola Mobility is in the best shape, but the Google acquisition casts a spell of uncertainty over their future, RIM and Foxconn are both on shakey ground. There have been engineering layoffs at all of the companies creating a surplus of young engineering talent in the area. Unemployment and the cost of living are both high. No jobs for a 60 year old engineer. Time to leave.

Maybe not such a stupid show after all.

It's still a stupid show, and I watch it. It has been very educational, and several of my co-workers have realized that I was right, and now watch it too. It is also interesting to see how the moral standards of the succesful and popular "villains" has declined over the years.

It made me realize that I could not succeed in the group that I was in (part of the cell phone division) due to the way the game was being played. I lack the capability to stab a co-worker in the back while smiling in his face (my boss was quite capable). I left that group 12 years ago and have been far happier since. If I had stayed, I would have had a heart attack, or my boss would have a phone up his a$$, either way the outcome would not have been positive. That entire group was part of the 353 person layoff.

If George "Tubelab" is finalizing a design, I suppose that's up to him, but the theory of an optimized "Tubelab SE Power Supply" board with some extra board edge connectors to carry over B+, filament supplies, pilot light feed, etc

I already have a board designed. Like the Spud SE, it cost too much to produce in small batches using my current PC board vendor. The board I have is quite simple with a tube rectifier for main B+ using the existing conventional power transformer. It uses a second auxiliary power transformer to provide +/- 150 volts for bias and mosfet drain supplies. This removes the main B+ from the mosfets on the TSE board allowing them to run at higher current while staying much cooler. I'll look at making it fit one of Seeed's standard sizes and see about getting some made.

I have several other simple board designs in the works too.

A Tubelab board usually goes through an itterative design process. I usually home cook a single prototype to test the design. Then I will refine it, redo the layout and home cook another board. This loop may repeat several times until I am totally happy with the results. Then I spend $1200 to $1500 for a batch of 100 boards. I need to recover that outlay before doing another batch of boards. The TSE took almost 2 years. THe SSE took off, repaying everything in less than a year. I have just ordered the 4th batch and it is the first Version 2.0 batch. The SPP has broken even after about 2 years.

This process doesn't work for something that will not sell 40 to 50 boards a year. The Spud SE, the auxilliary power supply, and several others will not sell that many in 5 years! I have resisted using Chinese PC board vendors, but maybe this will work for small volume runners.

The existing Spud SE board is 4 X 6 inches. I am sure that it can be squeezed into 100 X 150 MM. I'll need to check the 9 pin sockets and adjust the PCB pattern to fit.

What software did you use? Looks like the service likes a Cadsoft Eagle Design Rules check.

I have been designing PC boards since the days of tape and mylar. About 20 years ago I did some consulting for some guys that wanted a complex design done using a mix of analog, digital, and RF circuitry done. I explained that tape and mylar wouldn't cut it. They bought me the full DOS version of Eagle. I have been using it ever since, paying the reasonable $100 to $200 upgrade fee every few years.

Cadsoft was recently acquired by Newark / Farnell. The released a new version and now want $650 for the upgrade! This does not fit my budget, so I will continue to use the old version for now, while looking for other options.

Allied/RS has acquired DesignSpark and are giving it away for free even for commercial use! This is an obvious attack on Eagle's market dominance, and could not come at a better time. It also happens to read Eagle libraries.

I have downloaded and installed it, but haven't had the time to learn it. It is not as sophisticated as Eagle but is probably sufficient for our needs.

[Fenris]

Originally Posted by squiffiness

That's a pretty amazing price, all things considered. In what units (mils or mm) did you design your PCB's?

What software did you use? Looks like the service likes a Cadsoft Eagle Design Rules check. Since I am new to all this, I'm trying FreePCB. It has a design rules check, but I am too novice to know whether it's check is consistent with the Eagle check AND whether the Gerber files it will produce are a format compatible with your fab service.

I use FreePCB as well and generally use mils. I'll switch to mm to design a part for the library that is in mm or to make a board outline that has to be in mm, but I think in mils so that's what I use. Free PCB has a steep learning curve and doesn't have all the features of Eagle, but it works on low end machines and doesn't need to be installed on a host computer (I used to run it off a thumbdrive for portability, now I run it from Dropbox so I can use it anywhere on any machine). I don't really use a DRC, but like George said, tube layouts are pretty simple. My eyeballs are my DRC, I just make sure I connect everything up, don't cross the wires, leave enough space between traces, and usually end up iterating the design a few times until I get the least compromised design. To make the Gerber files, just go to File...Generate CAM Files... and then check the layers you need. A free Gerber viewer will help you to see what was outputed, but I've had no problems with the default settings. The Seed Studio PCB service requires you to rename the files into their format, but that's no big deal.

[squiffiness]

Originally Posted by tubelab.com

We had 353 people laid off in a single day.

All I can say to that is WOW!

Originally Posted by tubelab.com

...RIM and Foxconn are both on shakey ground.

Research In Motion is a major employer in my college town (Waterloo). The manufacturing sector in Ontario, Canada, heavily dependent on industrial and automotive parts fabrication, has seen some nasty downside in recent years. Unless one wants to move to Calgary and join the dirty oil crowd in the tar-sands, or take work at a fraction of what they're worth, good luck and enjoy asking, "would you like fries with that?" or "here's you're pizza, that'll be $15.50 please"

Originally Posted by tubelab.com

It's still a stupid show, and I watch it. It has been very educational...

Yep, just like every corporation I've ever worked at...you've got the crazies, the loyalty wonks, the delusional, the criers, the connivers, the politicians, the diplomats...

Originally Posted by tubelab.com

I already have a board designed. Like the Spud SE...

What I was driving at was simply to take the P/S components on the SimpleSpud board off to a separate board to squeeze the main board size down to 100mm by 100mm to get the overall fab price down.

Originally Posted by tubelab.com

The existing Spud SE board is 4 X 6 inches. I am sure that it can be squeezed into 100 X 150 MM. I'll need to check the 9 pin sockets and adjust the PCB pattern to fit.

I posted a FreePCB library footprint file (.fpl format) and a calculator (Excel) in an earlier post if that will help.

Originally Posted by tubelab.com

Allied/RS has acquired DesignSpark and are giving it away for free even for commercial use!

I have it downloaded and will give it a try too, just for giggles.

[MellowTone]

Originally Posted by tubelab.com

In post #38 Fenris pointed out a cheap PC board vendor that might make this project worth reviving.

*snip*

If I can really get 10 boards for $70 or 50 for $160 this little board may be worth doing. I will have to make a PC board library part for the 6LR8 socket and update the board. I dont think a dual footprint board will work because the two pin circles will overlap. I like to use oversize pads for all of my boards because many builders have never soldered a PC board before. The large pads make it almost impossible to destroy a board. I already have all the parts(25 sets) for populating the boards including the tubes (pre tested and gain matched). They have been in my closet for a few years!

Sounds good. I'd be inclined to go for a dual socket design (ie. two Compactron sockets directly behind the Novar sockets wired up to the board). Now IIRC that should be okay if the leads are kept short enough (I was thinking around 2 to 3") although I might be concerned about high frequency oscillation. I would be able to put a grid stopper on the Compactron socket as well to overcome this I would think?

Originally Posted by tubelab.com

I work full time as an engineer in a Motorola plant. I am one of about 20 of the "old timers" left in a world of young people. We must be very careful to follow the rules to avoid being replaced. About a year ago is was made very clear that computer use is monitored and anything related to Tubelab is a violation. I can't open Tubelab email from work. Limited forum reading is permissible during lunch. I can post simple replies from my smartphone (they can't monitor that). Most days I get home from work around 7PM so I get to read email on weekends, and can only answer some of it. I try to cover all the support issues from customers first, and sometimes that's all that gets done.

That was by no means a jab at you, I understand the situation you're in, been a rough couple of years. Seems more and more common for companies to monitor employees like crazy these days. They never quite seem to realise that at the same time they're destroying morale and stressing their employees to the edge. I'm just glad I get along well with management where I work, it really makes a difference. Sure, fear is one way to get results, but you get better results when your employees actually enjoy coming to work. But I digress.

At the moment, would you be expecting the run the same iron as the original Spud? 270-0-270 power tx and 5Kohm output?

[squiffiness]

Originally Posted by MellowTone

Sounds good. I'd be inclined to go for a dual socket design (ie. two Compactron sockets directly behind the Novar sockets wired up to the board). Now IIRC that should be okay if the leads are kept short enough (I was thinking around 2 to 3") although I might be concerned about high frequency oscillation. I would be able to put a grid stopper on the Compactron socket as well to overcome this I would think?

The service that Fenris suggests can make 50mm by 50mm two layer boards for $10. Why don't we end users just design our own daughterboard adapter? You know, a 9-12 pin or 12-9 pin and you can place any stoppers or adapters, bias resistors/diodes, etc. on that board.

The pin sizes are known (.050" and .040"), but it sure sounds like the sockets that are PC mount will vary to the point where you'll need to know the source and bought some to confirm the PCB pad circle diameter for your socket.

Originally Posted by MellowTone

270-0-270 power tx and 5Kohm output?

Was the goal a B+ of 300V? I seem to remember reading that the B+ in the proto-board sailed a tad high at 320V.

I did a PSUII simulation on the power supply on the schematic and found that a 250-0-250 HV supply (@200mA), the bridge rectifier, followed by a 100uF, then the 150R, then a 150uF cap produced about 300V on the dot with a load estimate of approx 120-150mA, based on about 10mA for the triode sections and about 50-60mA for the pentode sections.

[tubelab.com]

I'd be inclined to go for a dual socket design

Any way you lay it out a dual socket board will be bigger, increasing the cost. I haven't looked into it, but dual socket designs are usually best done with the common parts in between and the sockets on the edge of the board.

At the moment, would you be expecting the run the same iron as the original Spud? 270-0-270 power tx and 5Kohm output?

I would do a bit more testing, but that is a good safe combination. I had that board on a power supply and cranked it pretty high. 320 volts and 5K ohm is good, maybe a bit more voltage.......