• Disclaimer: This Vendor's Forum is a paid-for commercial area. Unlike the rest of diyAudio, the Vendor has complete control of what may or may not be posted in this forum. If you wish to discuss technical matters outside the bounds of what is permitted by the Vendor, please use the non-commercial areas of diyAudio to do so.

UNSET is coming?

There are currently two prototype UNSET boards in existence. They are identical except for the jumper setup so that any octal tube can be used. Both have been in operation off and on for over a year.

If the design was deemed ready to become a Tubelab product a clean up of the board is needed. There is currently no silk screen or solder mask layers on my boards since I made them myself. All of the reference designators were automatically assigned by my old version of Eagle, and they need to be manually changed before sending the board out, regardless of any layout changes.

The only issue I see right now is heat build up. Unfortunately is may be a big problem at the 15 WPC power level. A better IR thermometer is on order.

I have been testing the prototype amp seen in post #155. At the 15 WPC power level mentioned in that post the heat sinks for the cathode drive mosfets reach about 65 degrees C after about 20 minutes of operation sitting on the bench with zero airflow.

Most builders will choose to put the mosfets and their heat sinks under the top plate as is usually done on the TSE-II. I believe that thermal issues may arise in this situation. The best path to determine if it is a real problem would be to build a complete amp in a chassis. It might take some time to pull that off under my current situation.

There are several ways to improve the thermals, but most mean major changes to the current board layout.

This board design is the same size as the TSE-II but contains more parts and the mosfets dissipate more power than those in the TSE-II. This component density leaves less room for airflow holes in the board. The mosfets burn about 6 watts each at the 15 WPC level. I will measure the mosfet temp at several power levels when the new thermometer arrives.


1) The first possibility is to test the existing board with the mosfets and their heat sinks moved off board on short wires. This might work and would not involve major board work. I am hoping to try this in the next few days.

2) The mosfets could be relocated to the rear edge of the board to make heat sink or external mosfet mounting a bit easier. This is nearly a new board design.

3) At 15 WPC the rectifier tube is at its limits. It is also burning 12 to 15 watts of power. Some builders desire a a tube rectifier, some don't care. The current design uses a 5AR4 with diodes in series with each plate. The tube can be removed and jumpered across to create a solid state only solution. I ran my board without a rectifier tube during most of the time I used it with speakers last year. I tend to prefer the dynamics it has. If a rectifier tube is not deemed a necessary feature, I can move parts around to use its board space to better shed heat. This would also be a pretty major board change.

4) Do nothing and find the maximum safe power output available in the current board design.

These issues apply to the current "all on one board" design only. It is the only UNSET design that could be ready for production in a short time frame. Many people have expressed the desire for a modular solution.

I have always avoided a modular amp design due to the added complexity at build time. Having to interconnect several boards leads to more opportunity for error, and requires better build documentation. I spent much of my early engineering career at Motorola "un-modularizing" two way-radio designs to improve product quality. Nevertheless, a modular design does open up several options that are not available any other way.

The setup, NRE, and per board costs at low volume with my previous board house made small board runs uneconomical. That's why the current UD board is $25, and I still needed to buy 50 at a time. That issue is now gone opening up more options.

I have been talking about building a 1 KW vacuum tube amp for nearly 14 years, but have not done it yet. I got the OPT's in post #7 here:

Eight 807's in Push Pull?

I need to build it before I get too old to move it. This must be a modular design and the CED / UNSET technology is likely the way to get this done. Therefore there will be a modular UNSET solution capable of high power levels. 1 KW, no, not in any SE amp without it's own power station and water cooling. The BIG ONE will be P-P, but we have already seen an UNSET board do over 250 watts in push pull.

To make the BIG ONE, a driver board similar to the current UD board will be "designed," or just lifted right out of the current UNSET proto board. There will be an output board, also lifted right out of the current UNSET board (2 or 3 needed per channel for 500 WPC), and a more universal power supply.

These boards do not require much new design, since I have already simulated and tested the circuitry on the existing UNSET, and the older 50 to 80 WPC push pull board. I see two new module designs needed for the BIG ONE. One would be a phase inverter board for running the UNSET, the SSE, and maybe even the TSE-II as a push pull amp. The other would be a microprocessor controller / watchdog to keep tabs on a 500 WPC amp. Things can go very wrong, very quickly at this power level. My toroidal OPT's will need auto biasing, and maybe auto matching to control DC offsets that can cause saturation in the OPT. Saturation in a big amp can and WILL blow stuff up.

Now, any ideas for that 625 volt, 3 amp power supply to feed this beast? A dumm blonde power supply experiment is under way.
 
Rushing to get something out there that doesn't appeal to many people will only end badly for most users, and Tubelab. When it was time for the original TSE to be replaced by the TSE-II, I collected a long list of possible additions, upgrades, and fixes for problems like too much heat build up seen in some original TSE builds. Most of that list was incorporated in the TSE-II which has been quite successful.

I would like to make the "all in one" UNSET board something useful to the widest possible audience. I have been thinking, and re-thinking how to best do this, and I came up with yet another idea last night when I couldn't sleep.

I have sent my boards to every continent on the planet except Antarctica. So far they have all used tubes that are available everywhere with the possible exception of the 5842 in the TSE-II. The situation gets a bit more sticky with TV sweep tubes. I had originally thought of doing two or maybe three different boards to handle octals, 12 pin Compactrons, and 9 pin Novars and Magnovals. That leaves out the 9 pin miniatures that can be used for low powered amps like a headphone amp. All would use the usual 5 volt octal rectifier like the 5AR4 / 5U4 , which limit the power output to about 15 WPC.

So far the current proto board works great except for the heat, but the 5AR4 limits the power to about 15 watts. 40 WPC has been seen with an external power supply and one glowing output tube per channel. The board is capable of 40WPC, but a single sweep tube per channel is not.

Several different boards would be required to cover all possible tube choices. The mosfet / heat sink is still located in the middle of the board. I propose another more generic way of doing this that could make one board fit everybody's needs.

I propose one board that contains a pair of driver tubes, one for each channel, and all of the discrete parts for the complete amp. The three mosfets (two drivers and a screen regulator) would be along the rear edge of the board, so that they could be board mounted yet heat sinked to the chassis, real panel, or short wires run to an external heat sink.

The output tubes would be mounted off board with wires running from the board to the tube sockets. This would allow the use of ANY output tube from tiny 7 and 9 pin miniatures, to big fat transmitter tubes like the 4D32. It would also make parallel SE operation with two or three tubes per channel in the 40 WPC range possible.

The board would incorporate provisions for solid state rectification, or off board rectifier tube or tubes. This way the 15 watt barrier could be broken with damper diodes or two 5AR4's.

Any comments?
 
George, I think that is an excellent idea. The somewhat crowded layout and heat buildup were the two chief concerns that kept me from buying one of the original TSE boards for years. I purchased the newer board once it was clear those issues (along with the obsolete regulator) had been addressed.

BUT... immediately prior to the announcement of its obsolescence I was seriously considering a TSE build with the outputs mounted off-board. This fits right in as I have loads of possibilities lurking in my stash.


This is exciting, indeed. :)
 
Member
Joined 2004
Paid Member
I would like to make the "all in one" UNSET board something useful to the widest possible audience. I have been thinking, and re-thinking how to best do this, and I came up with yet another idea last night when I couldn't sleep.....

Any comments?

Hi George,

I admit that I have been overly vocal! And I apologize again for my impatience.

But this is great. I’m not particularly vested in the exact outcome, because I think you could adjust as the demand (for boards) lead you.

In general, I think an option for SS rectification is great. And please allow for the “big xLW6” tubes. That’s all I want for my Christmas this year.
 
All Sockets "Off-Board"

Personally, I like the idea of a power/interconnects board, as it gives max flexibility in both layout and tube pin-out.

I previously cobbled together a couple of "power drive" proto boards myself (to test the concept before I committed to the TSE-II monoblocks) which included the driver CCS, the coupling cap, and the MOSFET with resistor "network". Everything else was point-to-point. Worked surprisingly well, although it was pretty sloppy at the end, what with all the "edits". Also, I used a Pete Millett regulated fils supply, so I didn't need that on the board.

For example, the board could include:

B+ P/S filters (like SSE and TSE)
CCS (ie 10M45)
MOSFETs and associated resistors/trimpots

Rectifier, drivers, and Power tubes would all be "off board."

If the board (or boards, if mono) had a narrow form factor (say 2-1/2" or 3" wide) they could be mounted VERTICALLY, which would be a lot better thermally.
That's how I mount the Millet regulated fils supply, BTW.

That is, all of the heat-sinked "sand" could be arranged in a row, ie diodes/fils regulator/CCS/ MOSFET with a row of holes in the chassis directly above.

The counter argument, of course, is that off-board sockets increases the risk of screw-ups, some of which might be spectacular.

In addition, off-board sockets make care in wiring more critical, especially for the driver(s) and the filament wiring (although DC makes fils wiring lots less fussy).

IF the UNSET is intended to be dead-nuts simple and a good "first build" (like the SSE, which I think is the best beginner tube amp out there), then I think that at LEAST the driver sockets must be on-board.

HOWEVER, if as I suspect, the UNSET requires more sophisticated build and adjustment techniques (like the TSE-II and the Universal Driver UD board), then just give us PADS for wires to grid/plate/anode/screen/fils and we can figure it out.

Alternately, screw terminals for the socket leads would make a super-flexible prototyping amp that could be ENCLOSED and maybe even dragged "upstairs" to try-out on the good speakers...

None of these ideas are critical; many of us (George included) have hacked the TubeLab boards within an inch of their lives. I still have an SSE board which has hosted 6AV5s, 6BQ6s, and assorted other mongrels. Others have "glued" DHTs on there.

Just some thoughts....
 
If the output tubes are going to be off board to allow flexibility, can the driver tubes be off board too?

The UNSET board uses a pentode driver with local feedback from plate to grid making it operate like a triode. The tubes that I have been using have a rather high Gm and are designed to operate at frequencies well into the VHF region. This makes layout somewhat critical to avoid oscillation. I'm sure that some builders could do a good job wiring up a socket to the board in a manner that would work, but just as many would be building radio frequency jamming devices.

It appears that a few builders would want something like the two boards I have been using. It works fine except for some possible thermal issues. The only way to quantify this is to build some amps. I have decided to get a small quantity of boards made like the two that I have, all parts on one board. I worked on board cleanup all day only to have my old Eagle 5.11 go off into the weeds and eat my work. I'll start over tomorrow. This board can use about 10 different driver tubes, and some of them are cheap.

The modular version will have a driver board with an even wider selection of suitable tubes
 
Sounds like a really good “hack” (I mean that in a good way, like “IKEA hack”) to avoid using those precious 5842 drivers. What sort of drive are we talking about, George?

BTW I’m good with driver tubes on the board; don’t relish using grounded coax to wire off-board driver sockets. Are we gonna need tube shields for our pentode? Just curious...
 
PS re: “thermal isssues”

I am using low-speed muffin fans in my TSE-II monoblocks, as they will live in a smallish cabinet tucked in a corner. Generally, not a good thermal formula. Once I have it sorted, will publish...but early returns say that the bias stays rock-steady with quite small CFM ( and therefore very low dBa fan noise)

Easily the best amps I have ever built, by far. Non-audiophiles comment on the sound: “ooo, that sounds good, what did you do?” A natural mate to my open-back, 6 dB 1st order cross-over 96 dB/W speakers.