After a 14 year run, the TSE must DIE!

[Tubelab_com]

Friends, the time is here. After 14 years virtually unchanged, the TSE as we know it must go away. Why?

In today's electronics market a semiconductor company introduces a new part, makes a test batch, prints up a data sheet, and announces the part. If it doesn't generate some immediate interest, it may die right there with no production parts ever being made.

If there are enough inquiries via the sales engineering force a production batch may get made. Sadly if the new part doesn't find it's way into a million unit volume product, or have a high enough profit margin to support a lower volume within a year or two, it will vanish from the market.

We (Motorola) spent a lot of time redesigning phones as parts would come and go. It was a big issue in products with a long life like police radios. Today's high volume consumer electronics have a two year or less product life. New phones come out in less than a year. With each new phone, new semiconductors arrive, and old ones die off. So what does this have to do with the TSE?

As we have seen with the mosfets, the old Toshiba 2SK2700 went away long ago, a victim of slow sales and ROHS, but newer and better mosfets have always appeared that drop in place. They tend to stay around for a year or two, then vanish.

Unfortunately, the Sharp filament regulator chip went extinct several years ago. It was a new part 15 years ago when the TSE was designed, and remained available for over 12 years, but went away when Sharp had financial difficulties. Sharp is now owned by Foxconn and is no longer a force in the semiconductor market.

It seems that the TSE builders were the only people using these parts, Mouser and Digikey both had ample stock, so things were still OK, and I quietly began looking for a suitable replacement part. There is NO DROP IN REPLACEMENT PART available.

I have tested some of the regulators that have similar specs, and so far find none that work as good as the old Sharp regulator.

Recently, almost overnight the regulators in stock at Mouser and Digikey vanished. I noticed that a third party reseller started selling these chips at about the same time. Did he simply buy them all and triple the price, or did the major distributors flush their inventory of slow moving parts, I don't know. The chips can be found here, I have never ordered from these guys, so I don't know what they really have or how they got them:

PQ5EV5

I have ZERO boards left in stock, and I have been including a Sharp regulator free of charge with each board purchase, and will continue to do so until these boards are gone.

My chips are from a Digikey purchase made in 2013 when I was preparing to offer parts kits for the TSE, SSE and SPP. Family and job issues killed the parts kits plans.

Once these boards are gone, I will likely not get any new boards. A redesign of the TSE is in the cards, but what do YOU want to see?

Some possibilities are:

1) a re-spin of the same design with a different regulator in place of the Sharp chip. This would use the same size board, footprint, and same chassis design.

2) a two board approach with the power supply moved off board to allow for more room on the amplifier board. The TSE was originally intended for 45's and some modern 300B tubes are nearly touching each other. Heat is another issue with the current design at high supply voltages.

This would be a clean slate design, be a different mechanical design and need a new chassis design. It could incorporate the same screw terminals that are seen in the SSE and SPP boards. Other circuit improvements could be incorporated, but the amplifier design is sound, and I see no major changes. The power supply could be improved, but is constrained by size in the current design. Two boards would obviously cost more than one board, but the board cost is not a major issue in a high end amp. The separate power supply board could lead to multiple supply choices for different build choices.

3) a larger one board approach with power supply and amp on a single board. Again, a clean slate design, but with more design restrictions than a two board approach. This may still be limited in size to fit the standard post office mailing supplies.

4) mono blocks. Again, and idea to minimize heat.

5) I'm open to suggestions.......what are YOUR ideas?


PS......there ARE some new amp designs on horizon, both P-P and SE....and then there is the UN-SET (currently still a proto)........stay tuned.

-----UPDATE-----3-18-21

The PC board vendor that I have been using for over 15 years has screwed up 3 out of my last 4 board orders. Orders are down, so I have ordered boards in smaller quantities driving the cost per board up. The SPP boards that I ordered in January never got made despite me receiving a confirmation number. I finally got them two days ago. A screw up with the last TSE-II board order has left me low on boards, and I ran out last week. I ordered more boards from a different board vendor that I have never used. They are due here in the middle of next week. I will need to build one and test it thoroughly before offering them for sale. The new boards arrived and have been tested. They work fine and are currently available.



-----UPDATE-----6-8-19

The new TSE-II board is done and available, although a full set of assembly instructions are not yet finished. I am posting the current BOM and BOM notes here so it will be easy to find.

-----UPDATE 6-13-19-----

Board schematic added. See post #335 for some info.

-----UPDATE 6-27-19-----

It is stated several places in this thread, but that gets lost in the hundreds of posts, so I'm putting it here too:

I received a PM asking about the cost of a TSE-II board. Due to changing to a new lower cost manufacturer after my current vendor screwed up 3 out of the last 4 orders, the TSE-II board is now $35 like most of my other boards.

......................IMPORTANT BOM CHANGE......................

The part number listed for R5 has proven to cause trouble and loss of negative bias in two customer's amps. It has been removed from the BOM pending testing and verification of a suitable replacement. See post #420 in this thread, and this other thread for details:

TSEII: all was going well until...

 

[mctavish]

I have honestly hacked up a few boards to simply use the driver system... Output tubes off-board and a different filament supply (coleman).

I would say Modular would be the way to go. For us diyers paying a bit more for flexability never seemed to be a problem. After all, it was never really a beginners project IMHO.

Regardless George, I am sure what ever you come up with will be great!

 

[zman01]

George,

I also support the idea of going modular. My thoughts are:

1) Keep Power supply board PCB, and make it configurable for single channel or dual channel support (keeping in mind mono blocks)

2) Make the boards single channel, so 2 boards will be needed for a stereo build. Also, I am assuming that the single channel PCBs will be smaller. If folks want to, they can go for mono blocks too by combining 1 single channel PCB + 1 power supply PCB

Smaller PCBs should be lower cost individually, and for a 45, 300B, or 2A3 project, a few $$ more for 3 x PCBs (assuming a 2 channel build) should be acceptable to the community.

Too bad you have to retire the design because of parts becoming EOL. However we expect that you shall be able to come up with an even better design if you put your mind to it, and time allows.

Wishing you all the best George, and look forward to seeing what other diyA members have to say about a new TSE. (Should we start calling it TMSE - Tubelab Modular SE? )

 

[mctavish]

It should be called the BORG project!! An unstoppable fleet of square boards that together make an unstoppable amp! Driver boards, power supply boards, combined to assimilate all they encounter in to the vacuum tube collective!!!

But seriously... I feel like a stand -alone power supply board would be very attractive to many, even as it stands on its own. Then a driver board with the power tubes off-board. Even the greenest newbies should be able to run 8 wires to a tube socket mounted on the chassis. I would like to reiterate, I would have no problem paying a bit more for flexibilty and I feel others would be ok with it too. Lets face it, the costs of transformers for a quality tube amp far outshadow the costs of a few pcb boards.....


Those wanting to build a quality amp should not skimp on the transformers, much less a few more $$ for quality boards produced by a compentent designer with decades of tube experience.... My 2 sense...

 

[zman01]

Good points mctavish.

 

[w5jag]

If redesign ...

Pads sized / spaced for screw terminals for the connections into / out of the board - the builder is free to use the terminals, or solder direct to the pads as he / she sees fit.

Pads for local plate to grid negative feedback.

Current metering off the cathode, rather than the anode.

Pads for tip jacks or similar probe receptacles at test points to make metering easier for folks that don't install permanent meters. These could also be used to hard wire in meters.

Maybe some type of buck / boost switcher supply for the filaments instead of a linear design around a regulator chip. I was really impressed with how easy it was to dial up a voltage and current with the cheap Tusotek dc-dc buck / boost converter modules for fooling with oddball tubes. That would also eliminate / ease some of the heat issues around the filament regulator. Audiophiles might not like that type of supply, but they don't seem to like DC on the filaments at all as far as I can tell ... I would think switch noise on the output, if any, could be taken out with a notch filter. Or put both types on the board, let folks choose which one to build out ...

edit: why not make the high voltage a buck / boost switcher also? One transformer for all variations of the amp. State of the art.

I don't have an opinion on whether or not the power supply needs to be off the main board. I would think only the filament supply needs its own board if you go that route.

Crazy idea - option for different power tube sockets ... octal for things like 6A3, and for using indirectly heated tubes like 6550, 6L6, etc. I'm not sure there is a universal mount for tube sockets, so you might need something like a hole in the board, where a smaller daughter pcb with the correct socket installed can be screwed down onto the main board. Pads on each board to solder up the connections. The anode pad needs a through hole or socket for a plate cap lead. That would also let folks use 46, 47, 807, etc. Elekit had something like this on one of their amps where the user could switch between 6AQ5 or 6V6.

If you allow for indirectly heated tubes why not a regulator for a screen supply? I have some three terminal high voltage regulators

https://www.mouser.com/ProductDetail/689-LR8N3-G

that I have been meaning to try in a screen supply for my SSE, but that other project has been taking all my free time ...

Win W5JAG

 

[Tubelab_com]

Pads sized / spaced for screw

That is pretty much a given for everything I have designed since the TSE. As I stated many years ago the TSE was never intended to be a kit, or construction project amp design. I made one (complete amp)....I sold it... repeat. After a while I couldn't make them fast enough, and most of my customers were technical people, so I got some boards made and wrote up some basic assembly instructions. The instructions were improved several times, and loaded onto a CD, then someone said why don't you put this on the web..........

In all that time, the board hasn't changed. I still have one of the original home cooked PC boards from those early amp builds. It still works too.

Current metering off the cathode, rather than the anode.

That would require two completely independent floating power supplies, one for each tube. Doable, but complicated.

 

[kevinkr]

Coleman regs are sold in pairs and would complement a redesigned TSE quite nicely. I stopped making my own and starting using these instead some years ago in my own amps. (The only thing I use in my stuff that is not my own design.)

 

[OldHector]

I recently bought a TSE board, and have been struggling to source some of the obsolete parts, though there are helpful sites listing equivalents.
Could you consider mailing me a regulator chip? I have fixed the resistors, and have PCB 300B sockets on the way from China, and also some quite exotic Lundahl OPTs, so am already quite committted to this project.
BTW, have just wired together my SPP, which was my first foray into this dream to build my own amp, and initial impressions were fantastic. My partner listened in awe to what she thought was a poorly produced Elton John album, and couldn’t believe the sound stage. So now it is not just a pointless hobby ;-)

 

[Evenharmonics]

Another vote for modular system.

Modular monoblock is even better.

 

[sunil]

Modular sounds good.

 

[wa4htz]

I am also going to vote for the modular approach. Separate 2 ch amp and power supply would be my first choice but mono amp boards would be fine too. Please make ordering the power supply board optional for those of us who might want to wire the supply point-to-point. This would allow the builder to chose their favorite rectifier tube(s) and filter caps.


Also, it would be great if the amp board could be set up to use either the 4 pin output tube socket or an 8 pin octal socket. I realize this might be difficult though.


You might also think about just setting up either power supply or amp board to use Rod Coleman's regulator for the filaments. I'm not that familiar with it but it seems very well liked by everyone who has written anything about it.


Just my $0.02. I'll be watching this thread because the TSE or New TSE is what I want to base my high end headphone amp on.


Ken

 

[Tubelab_com]

I believe that a one board amplifier design, or possibly two boards, amp and power supply, will be the path of choice for the majority of Tubelab builders. Therefore I plan on doing something along the lines of the current TSE. I still don't know exactly what that will be yet, or exactly when. As for the modular approach......

Tubelab Inc takes up a lot of time, and has never made more than $1000 in any year, and it's been far less than that in the last several years, with a loss in 4 of those years. That was not a major issue while I was working full time and getting a fat paycheck.

My 41 year engineering career at Motorola ended in 2014, and I could no longer afford to subsidize the existence of Tubelab on a fixed retirement income. Sales kept declining as the economy slowed down and I made the decision to pull the plug at the end of 2016. Fortunately things changed, and orders started returning at the end of 2016, so Tubelab still exists. It has paid for itself in 2017, and 2018 looks like it will be close to even, so I'm still here, but there isn't much budget for new stuff and PC board spins.

For the near future I have taken on a contract engineering job to raise some operating capital. This means that there will be little vacuum tube work until it is finished. In the near term funds will be used to resupply the stock of the current PC boards.

Tubelab's sales, like that of others in the niche and DIY audio markets has always been cyclical, with peak sales coinciding with the Northern Hemisphere's winter months, and often zero sales in July.

Q1 of 2019 will determine whether I reinvest the contract engineering funds into more vacuum tube products, or venture into something a little different, or a lot different, continue contract engineering in the RF world, or maybe design a killer ham radio (I'm KB4LRE). My RF lab is nearly complete, and I'm no longer bound by Motorola's employment agreement preventing it.


.....From the above posts it seems that there is quite a bit of interest in the modular approach. When I first talked about a modular approach to amplifier design, AKA "Amplifier Building Blocks" there was virtually no interest. This was however about 10 years ago. I designed a Universal Driver board for a Push Pull amplifier, made three different test boards, then got really busy squeezing the life out of my dying engineering career. The driver board, and most of the tube stuff lay dormant.

In 2014 I left Motorola, sold, gave away, or trashed much of the "stuff" that I had accumulated throughout my life, packed up what was left and moved it all 1200 miles north into temporary housing while my retirement home was being built. During the year of living out of cardboard boxes, I took one of the best driver board designs, polished it up and sent it to a board house and had 35 boards made. Why????? I don't know, boredom maybe. They came back, looked neat, nice and shiny......then sat on a shelf, ignored.

In 2015 I moved into the new house, and set up a workbench. I got one of those boards out of it's package and build it. It worked, so I built another. There were several random tests over the next two years. Test boards for output boards in several flavors were made, 9 pin miniature, octal, 12 pin compactron, and 9 pin Magnoval. Some test amps were made. Why am I explaining this?

These are the first few "modules" or Amplifier Building Blocks" In that modular system that seems to be interesting now. That package of 35 driver boards? All gone, I built 4, saved one to replace one I fired, and sent the last two boards out yesterday. There were no instructions, no mention of these on the web site, nothing but this thread:

Tubelab Universal Driver Board, 2015 version

And this mediocre video shot entirely with my cheap cell phone:

YouTube

I ordered 50 more of these driver boards.

It takes some testing and a complete re-layout to turn one of the output boards into a design worthy of being a Tubelab product. This takes time, since I will do at least two test board passes before ordering a batch of production boards. There are two more Push Pull driver test boards that are operational, but not fully characterized and another that's still in "furball" (flying parts) state. The existing output boards can be used for SE (stereo pair) or P-P amps, but need to be made bigger. Boards for 4 pin (45, 300B) and 5 pin (307A) tubes need to be designed and made. It's pretty easy since the pinouts are common and no jumpers are needed.

There will need to be a few similar boards designed for use as SE drivers, and a few power supplies ranging from simple to fully regulated.

Maybe a complete amp strip, driver and output without power supply (choose one from the "building blocks" or make your own. Heater supplies, again, chose a "block" use a Coleman, or make your own. Want to use a driver block and make your own output board, or wire it point to point, go for it.

1st picture, several driver test boards, one octal output board, and one of the new push pull Amplifier Building Block drivers.

2nd pic, One of the driver blocks feeding a pair of 25L6's mounted on Turret Boards from AES. One of the 12 pin Compactron output blocks can be seen at the top left.

3rd pic, same block warming up some medium size sweep tubes.

4th, stereo, some nicer OPT's and a whole box full of used 6V6GT's.

5th, some 6DQ6 sweep tubes in stereo make for a rocking loud 100 WPC.

6th, a pair of octal output boards. They can be used for about 25 different tube types depending on jumpers and parts placement. These are set up for 36LW6's and pound out 160 WPC.

7th, a test board that accepts 7 pin or 9 pin miniature tubes. How much power can you squeeze through a 50C5 radio tube?

Any thoughts?

 

[tomchr]

Not to be a party pooper, but I'd recommend against the modular approach. Inventory management becomes a pain rather quickly that way. My DG300B consists of four circuits: A power supply board, a driver board, a B+ regulator board, and three filament regulator boards. Maintaining inventory in the right amounts of the various circuits is a pain.

Just saying...

If the main issue with the TSE is that the Sharp IC has gone away, I'd look at the LMZ-series or LM2267x-series from National/TI. They're switching regulators, so there's very little heat to worry about and you could place all the components right at the tube socket for minimal voltage drop. There're probably some newer options that could be interesting as well.

Tom

 

[Tubelab_com]

Inventory management becomes a pain

I was part of the industrial revolution in a factory that hand stuffed PC boards for two way radios, then hand assembled the boards into product, each of which was a custom order since each had to have individual crystals and tunings for every different police department in the customer base. There were over 100 different PC boards, and several hundred different piece parts that had to be hand built to make each HT-220 police walkie talkie. It took a week to pre-assemble the average walkie talkie, another two days in final test and assembly, and several more days for QC, audit, and packing, then another few days to get them into the UPS truck.

When the factory left the USA parts were loaded into an automated line, the GO button was pushed and 10,000 Nextel Phone boards were done by the day's end.

When I joined the phone team in 1999, I was asking complicated questions that had no answers, like how much does it cost you to place one part on the board. If we can replace 12 parts with one part that costs a bit more than all 12, you need that number to make the decision, since it improves the throughput of the line....

....the biggest question I popped, and the REAL EYEOPENER, "how much does it cost your factory to ship an empty box?" The factory had evolved from a system that sold and shipped $2000 walkie talkies. It was now shipping $100 phones, the cost of individual boxes in individual wrappers does not make sense when you have only ONE customer and he wants 10,000 units at a time! BULK PACK!!!!!!

I started my career as one of the hand assemble and tune guys. This was in 1973 and the only "computer" in the building was the IBM mainframe which printed out the paper "pick" and "build" sheets for each order. They were attached to the plastic tote for each radio. All the subassemblies and piece parts for each radio were then placed into the plastic tote which would travel down the final build line to a technician like me who would tune, test and assemble the final product. All tracking was done on paper. Mistakes were many and also hand tracked. Products had an associated quality number called DPU, defects per unit which for most products was greater than one, and for complex versions of the UHF HT220 was in the 3 to 5 range.

I would leave the line in 1974 to join the test equipment calibration lab where my job was to optimize that factory, improve the yields, and increase throughput. I was in that job for 10 years. During that time the entire factory was torn down and rebuilt 4 or 5 times, once moving to a new much larger building. MODULAR radios were introduced, which reduced the number of individual kits (PC boards, plastic and metal parts), at the expense of making, warehousing, and tracking the individual modules. This reduced the DPU to well below 1.0, eventually replacing the "DPU" bench mark with "DPHU" (hundred units). Mixed SMD and through hole radios were introduced, again reducing the number of kits, and improving the yield and throughput, ATE (automated test equipment) replaced a lot of the manual testing. DPHU was now under 1.0.

Obviously having too many different parts to track leads to problems, and increased costs. The flip side of this is the one board for EVERYTHING, and the differences are in the software. This also leads to problems. Let's just say it's well known that you can buy the cheapest model of a certain DSO, download the code for the most expensive model, reflash the cheap one and save yourself $1500!

There certainly is a happy medium somewhere, and it differs widely depending on your business model and sales rate. I have averaged 135 boards sold per year with several years below 100, and none over 200. It really doesn't make much difference whether I stock 4 different boards, or 10 different boards at my sales volume. Inventory management is by eyeball. All boards are in clear Tupperware containers on the shelf in plain sight. When they get low, I order more. Simple system, but it works.

At one time we offered parts kits for the boards. That may be an option in the future, but it does increase the complexity, and opportunities for error. The SPP board can really only be built one way, with two different output tubes, so we added the extra resistor needed for both tube options.....only one parts kit.

The SSE is pretty much the same. We had made parts kits for those once, and they were also not a big deal. The TSE however gets more complicated due to the wide variety in possible operating voltages, and the multiple driver boards would make matters worse.

I have often been asked why I don't offer a complete kit of all parts including tubes and transformers. It is not possible to stock all the possible combinations of transformers given the wide variety of operating voltages and OPT quality levels needed. It would also incur two sets of shipping charges for the transformers, once to get them to me, then again to send them out.

The only way that would make sense would be for a low cost minimal build, sort of a one size fits all "Model T" (all exactly the same) amp. That was going to happen in 2009, but two terminally ill parents, Sherri losing her job, and the beginning of Motorola's undoing, slowed my progress, and your copy of my design left me with several hundred unsold tubes, transformers, and parts. This created the biggest financial loss ever for Tubelab.

Any new ventures / boards / kits etc will be self funded in advance, parts kits may not return, and would only be viable for stable proven designs. New boards will only be ordered once there are funds in my account to pay for them in advance, and of course the really new stuff may be hinted at, but NOT divulged until I have boards in hand ready to sell.

OH, yeah....UN-SET and the CED are coming, but delayed until I finish the contract job so I can fund everything at once.

 

[tomchr]

....the biggest question I popped, and the REAL EYEOPENER, "how much does it cost your factory to ship an empty box?"

I buy boxes from U-line by the bundle to get the cost down. Even at that volume, the cost of shipping materials quickly gets to the $3-5 range for a medium sized box. Think about that next time you get "free shipping".

At one time we offered parts kits for the boards. That may be an option in the future, but it does increase the complexity, and opportunities for error.

I generally find that a link to a BOM set up with Mouser works quite well. Let Mouser worry about stuffing itty-bitty parts into itty-bitty bags. Some do request a full kit, so I order the parts from Mouser, add a reasonable fee for my time, and ship the goods together.
If you do go the kit route, Digikey will pack components into kits for you. They do charge for the service, but I seem to recall that the fee is somewhat reasonable, in particular if you order many kits.

[...] left me with several hundred unsold tubes, transformers, and parts. This created the biggest financial loss ever for Tubelab.

Yeah... I've never understood the "fill the warehouse" strategy. I see many go that route and the cost of storage ultimately ends up being a significant strain on their business. I obviously understand the economics of scale, but I've bought way too many parts in the past that I ended up not using and ultimately throwing out.

The other aspect of this is that fewer are building vacuum tube circuits. I sold about 8-10 DG300B board sets this year and a similar number of Novar Spud boards. Nothing to write home about, really.

[...] the really new stuff may be hinted at, but NOT divulged until I have boards in hand ready to sell.

That's always a good idea. I tend to get myself in trouble by disclosing too much too early.

Tom

 

[Jpeg]

I do have a request. I especially enjoyed the sound of the TSE with 2a3 tubes, but even with a very large heat sink and a separate 10 amp filament transformer, I never was able to get mine to work reliably with 2a3s. Can the new one have an option to be configured to handle 8 amps of 2.5 volts filament power? Just curious, are you sticking wish the 5842 driver tube or moving to something more readily available? I am fine with either a large board or modular boards. Hmmm looks like my TSE may be getting a new chassis. The old girl can use some love anyway.

Thanks George for creating such a great sounding amplifier, and making it available for such a reasonable price! Merry Christmas to you and yours.

Thanks,
Josh

 

[sherpa]

Just sent a paypal for a TSE board. I hope there is at least 1 left. Thanks George, i'm looking forward to what you come up with next for the 45.
Paul

 

[Mark Johnson]

I buy boxes from U-line by the bundle to get the cost down. Even at that volume, the cost of shipping materials quickly gets to the $3-5 range for a medium sized box.

US shippers can use the "flat rate boxes" which the US Postal service gives away for free, when you promise to ship by Priority Mail. This isn't a huge burden because the USPS option that's cheaper than Priority, called First Class, is only available on parcels less than 13 ounces. As the name implies, you can ship anything that fits in the Flat Rate box for the same price, regardless of weight. A couple lead bricks? No problem, same price.

They give away four different sizes of Flat Rate Boxes, at four different shipping fees. The sizes are Small, Medium, ADifferentMedium, and Large. Honest.

 

[rayma]

you can ship anything that fits in the Flat Rate box for the same price, regardless
of weight. A couple lead bricks? No problem, same price.

The maximum weight is listed as 70 pounds, but still pretty good.
The box would have to be reinforced with strapping tape for such
heavy weights, though.