• 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?

I discovered today that the SMPS that I have been using for the tube heaters is set to 28 volts. The tubes want 25.0 volts, so I readjusted it. This made some of the previous numbers a bit off, mostly in the balance between tube and mosfet power sharing, the max power numbers are still close, but some parts tweaking will be needed since the amp does not quite clip symmetrically any longer. The bottom starts to clip about a watt before the top. I had BIGGER things in mind, so........

I just had to do it.......

I removed the 500 volt filter cap, swapped out a 4.7K 1 watt resistor in the B+ line for a 10 K 2 watt since there is still a 500 volt cap down stream and hooked up the BIG power supply....the one that blows small parts into dust and makes sweep tubes explode.

I then cranked up the voltage and the audio oscillator at the same time. WHY?

Maximum tube dissipation happens at idle in a class A amp, lots of DC power going into the tube, zero audio power coming out. If I kept the drive level just below the point of clipping as I walked the DC voltage up, theoretically I could avoid melting a small tube when I shoved 68 watts into it.

68 watts into the tube gives me 30 watts out at 2% THD.....good enough.

That took 620 volts of B+ voltage, 23 of which were lost in the OPT. 522 volts were across the tube and 75 across the mosfet. Idle current was 130 mA. Power at 5% THD was 34 watts. That's 80.6 total watts input (520 volts @ 130 mA) for 34 watts out, 42.2% efficiency, not bad for a SE amp. My 45 amp runs about 20%, the 300B TSE-II about 30% and the power hungry 845 around 33%.

Good enough? I turned the power supply up a bit more, 640 volts brought 35 watts at 3% THD and 37 watts @ 5%. Now I need 90 watts in to get 37 watts out. Still 41%, but I'm going in the wrong direction.....All of this was done with the 50 cent 25DN6's, I need bigger tubes!
Early this morning I reconfigured the board for big tubes. And plugged in some 36LW6's and another big sweeper. I was surprised to find less power output. About 2 hours of tweaking left me with a dead board.
There seemed to be a brick wall in power output even at 650 volts and up to 160 mA. The original parts values from the simulation are still in the triodzation network. They were derived with a model for a 6DQ5 which is very much like the 25DN6's l started with. Once life is restored to the board I plan to try some 6DQ5's.

Unfortunately I don't have an LTspice model for any of the big sweepers so I may have to tune those in with so several pots.

I'm out of the house for the rest of the day so maybe tomorrow.
Late last night I fixed the dead board. One of those white ceramic 5 watt resistors that never go bad.....went bad. The mosfet that it feeds, however was still alive.

I poked around in the board with a scope probe and figured out where the brick wall of power is coming from. My feedback path (AKA triodization loop) is running out of headroom and clipping.....so some resistor value tweaks are needed. No big deal, except......

I decided that whenever playing around in a high powered amp running on a power supply that can shatter vacuum tubes, bad stuff can, and sometimes does, happen. I decided to remove the big fat sweep tubes and return to the mid size stuff and revert back to the old Fluke. With it parts don't usually explode, they just smoke or catch fire. Unfortunately one of the 36LW6's had become stuck in the tube socket. Attempts to extract it left the plastic base key stuck in the socket. I drilled a tiny hole in the PC board from the bottom and pushed the key out with a fat wire.....

....So the dumm blonde tries to glue the key back to the tube with some year old superglue and winds up with a vacuum tube glued to his hand! This leads me outside in the dark to fetch some acetone from the shed. I did manage to eventually free the tube and glue the key back on only to have it break off in the socket again this morning......

I now have the board operational again with the mid size tubes and am taking a short break from the clip leads and box of resistors experiments.
So the dumm blonde tries to glue the key back to the tube with some year old superglue and winds up with a vacuum tube glued to his hand! This leads me outside in the dark to fetch some acetone from the shed. I did manage to eventually free the tube and glue the key back on only to have it break off in the socket again this morning

You should seriously consider writing a book about your vacuum tube shenanigans. I'd buy a copy in a heartbeat.


Paid Member
2009-08-20 3:16 pm
KC Metro
....So the dumm blonde tries to glue the key back to the tube with some year old superglue and winds up with a vacuum tube glued to his hand!
Hah! The story of my life. :) I grew up with the stuff because my father was a lifelong R/C aircraft enthusiast. I recall my first introduction to cyanoacrylate adhesives was sometime in the '70's at a hobby shop in Fayetteville, NC. My dad was good friends with the owner, who showed him a miracle adhesive called "Zap" during one of our Saturday morning visits. They glued a quarter to the top of one of the glass display cases as a joke, and spent the next several hours watching various customers try in vain to pick it up - and giving up in puzzlement.

Of course the rest is history; my dad bought a boatload of the stuff, and I spent the remainder of my childhood gluing balsa wood to my fingers. :p

Not to make light of your situation as it is a bit unpleasant, but that did bring back some happy memories. BTW, the stuff lasts longer if you keep it in the fridge.

Watching your latest adventure with extreme interest while I finish my TSE-II build. I'm amazed by that little amp, and a bit embarrassed I didn't purchase a board sooner. If this thing sounds as good as you say it does, then I can assure you I won't wait 14 years to buy a board.
... tries to glue the key back to the tube with some year old superglue and winds up with a vacuum tube glued to his hand! ...

Dollar Tree has your back here - single use tubes of super glue - 4 packs for a buck. Also these neat organizers that are terrific for components. I use them as part of my tool kit / parts store that shuttles between the office and the lake house.


  • IMG_20190805_112148267.jpg
    129.2 KB · Views: 961
  • IMG_20190805_112716177.jpg
    164.1 KB · Views: 959
They glued a quarter to the top of one of the glass display cases as a joke, and spent the next several hours watching various customers try in vain to pick it up - and giving up in puzzlement.

I was the service dept tech at an Olson's Electronics store in 1971 and 1972. We had this stuff called Oneida Instant Weld. It came in a belt kit for fixing oddball Japanese tape decks, just a bunch of rubber belt stock and some instant weld (AKA super glue). I did exactly the same thing with a quarter, and it was still there when I left the store for a real job in late 1972.

One of the salesman one upped me by gluing the store manager to his chair and later squirting the stuff into the keyholes on his car......yes the manager was a jerk and the reason that all of the good people left and the #1 Olson's store in the country slid into extinction. That store was next door to the athletic office and frat house row for a little school called the University of Miami....an unlimited supply of rich kids with daddy's money.

single use tubes of super glue - 4 packs for a buck.

I had a couple of those quad packs....but they weren't in the tape / glue drawer when I needed them........

my father was a lifelong R/C aircraft enthusiast.

My only foray into that hobby was a Falcon 56 back in the late 60's. It took me months to build it, and hours to destroy it. The .29 engine it had wound up powering the Heathkit RC car...until I managed to destroy that too.

You should seriously consider writing a book about your vacuum tube shenanigans.

I doubt that will ever happen. I have fought ADD or ADHD my entire life....I can't even read a complete book, much less write one that has continuity. Fortunately all those articles in the magazines of the 60's that I read during my childhood just matched (or defined) my attention span, about 3 pages.

Now.....back to playing with electricity......
I stuffed a whole bunch of different tubes into the amp this morning, basically two or 4 of everything I had that had a pinout compatible with the 25DN6. Other than minor changes in the bias the results were similar.

The 25DN6 is rated for 15 watts max dissipation.
The 6CB5A is rated for 23 watts
The 6CD6 is rated for 20 watts
The 6DQ5 is rated for 24 watts

All of these that were made from the late 60's on have EXACTLY the SAME PLATE in them! This means that each should be capable of eating at least 26 watts, and they indeed can. Some of the older tubes that have their original small plates inside will melt at 26 watts. I had a pair of late production GE 6DQ5's up over 50 watts briefly, and they were rather unhappy and rather red faced though.

I have realized that Ohm's law is the brick wall that I have been banging my head into. I have a 5000 ohm OPT. In order to put 30 watts of audio into it, I must put 387 volts RMS across it's primary, or 547 volts peak. Even a sweep tube will saturate in the 50 to 70 volt range in triode mode, so now I need 600 to 620 volts of B+. That big Hammond was one of the lossiest OPT's I tested back when I got them. It gave up a watt to the EDCOR at 7 watts, so it's probably losing 2 or 3 at 30 watts. So it looks like 30 watts is about the limit for a 600 - 650 volt supply with a 5K OPT. I need to find my big Edcors and try one.

So How did I get my analyzer to show me 38 watts at 3% THD? I wired the Hammond to reflect 2500 ohms (8 ohm load on 16 ohm tap). Then I cranked the tube current up to 200 mA, at which point it got really mad, then I cranked up the drive. Yes I got 38 clean watts before I shut it off to cool. Time to try the big tubes again.


Paid Member
2009-08-20 3:16 pm
KC Metro
Just to be clear, you're using the Hammond 1628SEA transformers, right? I ask because I still have a pair of those left over from the only amp I built that I ever hated (a pitifully anemic 6J5/6550 design featured in a prominent DIY magazine). Maybe it's time to give them another chance

I've never hear of Olsen before, so I made a quick search and dug up this Olsen catalog from 1968. Man, what a trainwreck! It isn't so much a catalog as it is the whole advertising section from one of those old Popular Science magazines from the late 1940's. You know, the ones that were full of radio repair correspondence schools shouting slogans like, "I send you BIG KITS of parts!", etc.

Page 11 has a 20 lb "Surprise Kit", a "colossal bargain" for $5 USD. Just imagine - 20 lb of mail-order merde for today's equivalent of $36.59. And on page 21 is an entire linee of Wilesco steam engines (what the... ?) The 1974 catalog looks a little better (more upscale brands), but many of the speakers look cheaply made and appear to have those disgusting foam grilles.

I can see why you went elsewhere....
Before the attempt at flying model aircraft, I hung out at a local slot car track (if you can remember those). I rewound motors and even hacked up a couple of them and made one big one with a really long armature and 4 magnets. I won a lot of races and made a few bucks at that hobby, but that fad got too expensive for kids, and thus died off in the mid 60's. Before the track closed a Lafayette Radio Electronics store opened two stores down, which became my new hangout. They let me play the guitars and talk on the CB radio. When I finally got my ham license in the 70's a randomly assigned novice call was issued, KB4LRE. I liked it even though Lafayette was gone, so I still have it even though I'm now an extra.

Olson's was a step down from Radio Shack or Lafayette.....a pretty big step at that. They improved when they were bought out by Teledyne, at about the time Teledyne bought Acoustic Research, so we sold a lot of that stuff. Olson's house branded stuff was like the others built by the lowest Japanese bidder, most was junk, but there were some diamonds in the rough to be found. Their "Concord" branded cassette decks were made by Nakamichi and lived up to their standards. The potted OPT's were Tamuras, and all of the Olson branded guitar speakers were Utahs. Some of the house branded HiFi drivers were crap, but some were Eminence, and good ones at that.

you're using the Hammond 1628SEA transformers, right?

Yes, that's what I have. Tonight I successfully squeezed 40 watts through them at just over 3% THD. I stuck in a pair of the biggest sweep tubes I have, the 36LW6's that I used last night. This time I decided to beat ohms law with current, not voltage. I had to bend or break every spec in the book, but I was going to see 40 watts of SE power flow out of this thing, with both channels running, or blow it up trying......

I connected my 8 ohm load to the 16 ohm tap. This puts a 2500 ohm load on the tube. Mr Ohm says I need to put 316 volts RMS (447 V peak) of audio across a lossless OPT and it will draw 126 mA RMS (178mA peak) of current through the tube. This means that the idle current through the tube MUST be at least 178 mA to avoid touching cutoff. 178 mA of idle current will drop 30 volts of B+ in the OPT. The tube will see 70 to 100 volts across it on the current peaks of nearly twice 178 mA. This means I need about 550 - 575 volts of B+ to get 40 watts.

So I set the big power supply to 600 volts, set each tube to idle at 175 mA and cranked it up. I hit 43 watts at 5% THD, and 2.7% in one channel and 3.3% in the other at 40 watts.

there was about 500 volts across each tube at idle. The current was 175 mA that's 87.5 watts per tube, and you know what they were doing. The picture has had it's contrast and brightness enhanced. They were really not glowing that brightly, but one is obviously not even, and the other one is evenly glowing, is missing it's key, and has some permanent glue stain fingerprints, but most of the skin has burned off!

I reduced the voltage and current until there were no traces of glow in a dark room. That took 500 volts at 150 mA leaving about 60 - 65 watts burned in the tube and 32 watts of audio output. .

The Hammond OPT has a max current spec of 120 mA. I ran it at 175 mA so saturation was seen a lot. At 20 Hz I couldn't go past 4 watts, and at 40 Hz I could get 10. Even at the lower 150 mA current saturation was still a problem.

So if someone really wanted a 40 watt SE amp it would probably take a proper OPT, and TWO big sweep tubes per channel.

At my next opportunity I will revisit the 5K route with 30 watts as the target power level, and try the Edcor OPT.


  • P2920693_x.jpg
    670.4 KB · Views: 1,312
  • P2920694_x.jpg
    606.9 KB · Views: 1,332
  • P2920692_x.jpg
    614.3 KB · Views: 1,365
  • P2920697_x.jpg
    448.9 KB · Views: 1,367
Last edited:
Last night's experiments were somewhat a repeat of something I did about 13 years ago with different results. Big Hammond OPT VS big Edcor OPT.

Unfortunately, this time I don't have FFT capabilities yet (dead computer) and it is not a true A/B test since I simply swapped out one of the big Hammonds for an Edcor. This leaves an unwanted variable in the test, the fact that I'm using a different audio path for each OPT. I have already confirmed that the two channels are well matched in level, but the absolute distortion numbers for each channel are slightly different primarily due to different output tubes.

I would trust the power output and relative frequency response numbers to about 0.1dB, The absolute distortion numbers may be a bit off, but the relative changes across frequency are valid.

Previous testing showed a sizeable HF notch falling somewhere in the upper audio band on the Hammond. The depth and frequency of this notch was highly dependent on the tube and configuration being used. Much of my testing used a KT88 in triode or UL and showed a loss of 3 to 5 dB at 20 KHz. The big Edcor has the same notch, but is comes into play well above the audio range. The Hammond is considerably larger than the Edcor, and thus has more inductance and capcitance due to more wire and iron. All of this metal requires more energy to magnetize and demagnetize quickly, so it's losses will be higher, especially at higher frequencies.

The Hammond has always demonstrated monster bass, but subdued highs in listening tests, and the Edcor always made more power output in the same amp.

Almost all OPT's have this resonant notch, but it usually falls above the audio range. It is a resonant pole formed by the stray winding capacitance and the leakage inductance. Driving the OPT with a low impedance source is effectively like putting an equal valued resistor in parallel with this resonant pole. It lowers the circuit "Q" widening the notch and lessening it's sharpness.

I have preferred the Edcors in 300B TSE's and KT88 SSE amps which is why the Hammonds have been in their boxes for 13 years. My Edcors spent several years in a TSE that ran triode wired 307A's. I stole the power transformer and choke out of that amp several years ago, and eventually dismantled it before moving out of Florida, freeing up my early vintage Edcors which were painted bright metallic blue before Edcor switched to the more subdued color that used today.

I have decided that this little test board is going to get stuck into an amplifier ASAP. I want it to see some continuous use and have a lot of listening time with it since it is a rather radical departure form the norm. The original TSE went through a lot of changes in the two years before Tubelab was born. It was my first SE amp design, so maybe this one won't take so long to prove itself. So which OPT's will I use? Bench testing is fine, but it doesn't tell me which one I will like, so I will probably make a slab of wood amp like the first SSE, or drill the top plate for three different OPT's if I make a chassis.

Test Data......Test #1 both channels driven from the same audio generator. B+ is 550 volts, idle current is 120 mA, tubes were 36LW6's, OPT's wired for a 5K ohm load and no clipping was observed on either channel. Edcor on one side, Hammond on the other. I set the output power to an even number on the Hammond side, and measured the power on the Edcor side. The same HP8903 was used to measure power by moving the test lead. The Edcor consistently made about 12% more power than the Hammond. This matched the data from 2006.

...Hammond.......... Edcor
...1 watt...............1.125 watts
...5 watts..............5.611 watts
...10 watts............11.26 watts
...15 watts............16.83 watts
...20 watts............22.48 watts

Test #2 Frequency response. Note that most testers use a low power (often 1 watt, even on big amps) for this test to make better numbers. I ran each channel at 10 watts output measured at 1KHz, then varied the frequency recording relative power output and THD. Mild saturation was seen on the Edcor at 20 Hz and is the reason for the increased THD.

20 Hz.....-0.37dB 3.56%................-0.44dB 5.03%
30Hz......-0.17dB 2.12%................-0.22dB 2.77%
50Hz......-0.08dB 1.55%............... -0.09dB 1.81%
100Hz....-0.02dB 1.44%................-0.03dB 1.47%
1KHz...........0dB 1.20%.....................0dB 1.36%
5KHz......-0.11dB 1.01%...............-0.07dB 1.24%
10KHz....-0.25dB 0.94%..............-0.20dB 1.39%
15KHz....-0.49dB 1.24%...............-0.34dB 1.95%
20KHz....-0.79dB 1.89%...............-0.58dB 3.05%

This data is a radical departure from what I saw years ago. The "Hammond notch" is still there, but is moved safely out of the audio band. Tube current starts rising at the frequency extremes, which is normal.

OPT saturation and insufficient primary winding inductance causes increased current at low frequencies, and indeed the Edcor channel current starts rising below 30 HZ while the Hammond can run at 20 watts at 20 Hz.

Winding capacitance causes tube current to rise at the high frequency extremes. Tube current started increasing in the 15KHz range on both channels with the Hammond going up much faster. Neither channel had any problems cranking out 10 watts at 20 KHz which is probably more than will ever be seen in real music, and into the smoking tweeter range for many speakers.


  • P2920704_x.jpg
    839 KB · Views: 1,261
Last edited:


2013-09-08 4:40 pm
Hallo George, any progress ? been silent a long time in this thread, thought first you blown up, but see you live and answering other threads / topics ! =)

Will this project be of a near future with possible pcb:s ?

I would like to order from you but i still wait on the solution for the TSE-II About R5
Or it will work fine with replacement Ohmite 42J270E ?
been silent a long time in this thread

An unfortunate series of events (wife in hospital, grandkid in hospital, and myself with a broken tooth / crown / root canal) have kept the Tubelab workbench dark for several weeks.

Will this project be of a near future

I do plan for this to be a future Tubelab PCB, or more likely several PCB's, but it will not be near future. My wife was in the hospital for several days in bad shape. I spent a lot of that time in her hospital room with my laptop. I converted one of my simulations into a PC board, since I didn't have much else to do. That is the board seen in this thread. I had about one week of serious time to play with it, but it's hasn't seen power since.

Before it can become a Tubelab product it needs to undergo some serious testing and probably several board revisions. This board uncovered several layout mistakes (mostly part fitment issues) but did result in some serious surprises since I didn't expect it to work without major changes, much less work so well.

I now believe that a single board amp like the one presented here will be good to 15 or maybe 20 watts. Anything bigger needs more heat sinking on all of the silicon, so it will probably be divided up into multiple smaller boards. There could be a single board version somewhat similar to this one, and a set of modular boards for bigger SE amps and other useful things like a monster push pull version. Right now it's too soon to tell.

Before I go down this path I need to wrap up the TSE-II issues (like R5) and create some good build documentation, and update the web site. That is also true for the UD board.

Today I am in the process of clearing the workbench and setting up a space for shooting pictures and maybe video of board and amp builds and amp setup / testing. I have already put the UNSET in it's tote box on the shelf for now.

We had planned a two week vacation trip to Florida for mid September, but Sherri's health and hurricane Dorian may kill those plans. If we do go, I will take the laptop for some more board layout time.
Hope everyone is progressing on to a good recovery soon.

I've been on various meds for the last few years for high BP, with only modest improvement and annoying side effects. Then I discovered Olive leaf extract, which worked as good as the prescription stuff, so the meds got dropped. Then more recently I found out about mushrooms containing Ergothioneine, a new vitamin - soon. We have DNA and a blood transporter protein that specifically supports ergothioneine. Has been found to be DNA conserved across all mammals even (better than most human vitamins are) and is highly retained in the body by the kidneys, over a week half life compared to 30 minutes for most things consumed.

Mushrooms cause blood thinning (may not be related to Ergothioneine, don't know yet). The research article I saw on them recently suggested a half cup of mushrooms twice a week, I think. (found to clear up memory fog) Since I rarely consumed mushrooms in the past, I then went and over-did that with 2/3 cup of mushrooms every day, and developed bad nose bleeds for a couple of weeks. Cutting back to just 1 x 2 inch mushroom per day (using WalMart Baby Bella salad mushrooms), I've been clear of any nose bleeds, and my BP has dropped back to ideal territory. No more meds or olive leaf needed. Worth a try, check with doctor. Do not take mushrooms before any surgery (for weeks).

Some other common items can cause or aggravate nose bleeds too. Turmeric/curcumin, aspirin, Omega 3 oils, various allergies, so steer clear of those while steadily on mushrooms. Some mushroom types need cooking, or may cause allergies too. (especially more exotic types)
Last edited:
One caution
for consuming a mushroom every day (and no more than 1 !!! or maybe nose bleeds or too thin blood). Ergothioneine in mushrooms is good at removing Zinc from the blood, and could produce a Zinc deficiency over extended time. Best to take 1 or 2 RDA of Zinc a day with a steady diet of mushrooms, ie multi-vitamin/mineral (and no more than 3 x RDA, Zinc has a narrow tolerance window). Just to be safe.
This sounds really interesting, smoking amp! My father will be interested in this for sure. Why I'm polluting this wonderful thread is the following:
Most mushrooms have a so called adaptogen function: The immune system will use it as some sort of universal fuel and use what it thinks is the most urgent to do. Every mushroom is different. You will find lots of scientific information in englisch and other languages, as mushrooms have been used for thousands of years, and just recently get discovered by the "modern" western medicine.

I don't know if every mushroom thinnes the blood. There are hospitals using mushrooms in the US for different purposes.
selfhacked is a good starting point for 4 different mushrooms, as they have listed all the cited papers.
Sorry again for the off topic!
Hallo George, any progress ? been silent a long time in this thread

Wake up time. We have this virus thing going on and the TV is telling us that we need to stay inside. So what do I do when I'm stuck inside......build stuff.

As I was experimenting with UNSET last year I mad a list of all the problems, and a plan to fix them. I hacked up the only working UNSET test board to try my "fixes," documented the results, bounced back and forth between the simulator world and the real breadboard until they aligned.

The UNSET is an amplifier design that uses CED topology for its driver and output stages.

I was also working on the problems I faced with the little push pull version seen in post #20. The CED does a lot of things very well, but it makes a poor LTP. The push pull amp uses a pair of CED's for the LTP driver, and it just never worked right. I wound up with one of my Universal Driver boards wired into the push pull output stage and saw some real good results.

I decided to take a little diversionary detour to test some things that looked good in LT spice. That resulted in a little perf board test amp with a mosfet split load PI that squeezes 20 watts from a pair of lowly 50C5 radio tubes, without over stressing them. I let it run overnight at 20 watts and it was still running fine the next morning. The unbuilt section to the left of the three wired up tubes will be a guitar preamp. This whole test board should morph into a 20 watt guitar amp, with a little more of my attention.

Lessons learned from the diversionary experiments are being rolled back into the UNSET and Push Pull designs. Resulting in a new UNSET test board, seen here being populated along side the original UNSET board, and two days later, fully operational with the fattest sweep tubes on the planet, the 36LW6.

Note the revised jumper scheme for the output tubes. Anything that comes in an octal socket can be connected up, though obviously not all octals will work. A small daughter board for compactrons can be made too.

I realized that even a 5U4 is insufficient for these big tubes. I am using a bench supply with them and running 480 volts of B+ at 275 mA total power supply current for 20 WPC. This kind of power will probably need silicon diodes, but amps below 10 to 15 (maybe) WPC might get by with a tube rectifier if desired.

I have found some things that need changing, so at least one more test board and a lot of testing will be needed before the UNSET can become a Tubelab board.

I'm about halfway through some serious testing. The data will be posted as I compile it all, but so far things look good.


  • P3730052_x.jpg
    976.5 KB · Views: 317
  • P3730079_x.jpg
    865.2 KB · Views: 290
  • P3730080_x.jpg
    873 KB · Views: 279
  • P3730090_x.jpg
    680 KB · Views: 255