Gentlemen, these results are getting very good! Congratulations! This is encouraging that your are both getting similar results. Imaging the fanfare and accolades if a 300B SE amp sported these numbers!
I’m not clear what driver tube and output impedance George used during his reported measurements though. Didn’t George use the 1.5k primary hookup, while spiggs is still on 3k? George said he ran 135 ma while spiggs did his measurments at 105 ma, so they are at different operating conditions.
I’m not clear what driver tube and output impedance George used during his reported measurements though. Didn’t George use the 1.5k primary hookup, while spiggs is still on 3k? George said he ran 135 ma while spiggs did his measurments at 105 ma, so they are at different operating conditions.
Last edited:
Yes, my recent measurements were at 1.5K load and about 350 volts of B+ under full load. I'm guessing that better results would occur at somewhere around 380 volts of B+ and a 3K load, but I am looking to make a complete 20 WPC amp out of stuff I already have, and I have not found a suitable power transformer that will get me about 380 volts with a total load current in the 280 to 300 mA total or so range. Running the amp at 1500 ohms requires more current, but at a lower voltage. I have not tested my current setup at 3000 ohms yet, but I will soon. If I get 20+ watts with better measured specs and similar or better sound than what I heard last night, then I will use 3000 ohms for the amp build.
I have not yet found a combination with what I have available that will work with a 1.5K load so have mainly stuck with a 3K load. I recently tried using a AS-3T275 that I have but under load it only gave 318V of B+ bias at 150mA and performance was abysmal. I was using an adjustable regulator at one point to dial in different B+ values but managed to blow that up with the current UNSET setup so now only have 2 options using unregulated power, an Antek 3T350 or 3T275.
I have also tried using the OPT wired for a 6K load with interesting results. 1W THD drops down to 0.08% but the amp only puts out 10W. Considering the SSE that is my main amp is triode wired and puts out only 6W a 10W replacement would certainly be usable but I think we are looking for a bit more oomph here with the UNSET.
For myself listening tests have pointed me towards focusing on the THD spectra at different power levels rather than targeting the lowest overall distortion. My favorite combinations have had dominant 2H, if even slightly, almost all the way to max output. Using a 3K load THD spectra at 10W has been a good indicator, if I can get it dialed in to have more 2H than 3H at 10W the amp remains well behaved throughout the rest of the power range. I don't have a setup to use a frequency sweep for THD but when testing I do try the amp at a number of other frequencies and it generally shows the same THD spectra as at 1kHz. The current setup using 105mA bias behaves well in this respect. Listening tests involve a number of different tracks covering classic rock, punk, electronic, folk, classical, opera, jazz, blues and some experimental stuff that my wife struggles to identify as music.
I have also tried using the OPT wired for a 6K load with interesting results. 1W THD drops down to 0.08% but the amp only puts out 10W. Considering the SSE that is my main amp is triode wired and puts out only 6W a 10W replacement would certainly be usable but I think we are looking for a bit more oomph here with the UNSET.
For myself listening tests have pointed me towards focusing on the THD spectra at different power levels rather than targeting the lowest overall distortion. My favorite combinations have had dominant 2H, if even slightly, almost all the way to max output. Using a 3K load THD spectra at 10W has been a good indicator, if I can get it dialed in to have more 2H than 3H at 10W the amp remains well behaved throughout the rest of the power range. I don't have a setup to use a frequency sweep for THD but when testing I do try the amp at a number of other frequencies and it generally shows the same THD spectra as at 1kHz. The current setup using 105mA bias behaves well in this respect. Listening tests involve a number of different tracks covering classic rock, punk, electronic, folk, classical, opera, jazz, blues and some experimental stuff that my wife struggles to identify as music.
Last night I rewired the OPT's for 3000 ohms and swapped in the big Hammond power transformer. The B+ voltage under load is 425 volts. I found a usable bias point at 110 mA which results in 40.8 watts of dissipation in the tube (418 plate volts, 47 cathode volts). I then ran the usual tests. THD at 1 watt was 0.262% in the right channel and 0.238% in the left. THD at 5 watts was 0.549% in the right channel and 0.500% in the left. THD at 10 watts was 0.749% in the right channel and 0.720% in the left. THD at 20 watts was 1.43% in the right channel and 1.46% in the left. Clipping is seen at about 21 watts. Somehow I blew the input mosfet while hooking up the input pot and jacks for a listening test. No listening test happened.
I am fixing the board, and also plan to test the big Hammond 1628SEA OPT's at 2500 ohms. The lower impedance should bring a little more power, but the higher loses in the Hammonds will likely eat it all.
Since I had several of all the Tubelab board based amps, I had plenty of amps to choose from. Most of my listening was done via the PC which even had a turntable connected to it. I swapped out amps whenever the mood hit me. I sold, gave away, or parted out several of my amps when I had to move everything I owned 1200 miles on short notice.
I had three SSE's, a little version with triode wired 6V6's and $29 Edcor OPT's that made really nice sound, but was best used during quiet times since 2 WPC doesn't go far on 87 dB speakers. There was a SSE with EL34's in triode, and an SSE with KT88's in triode or UL for 14 WPC on 100 mA per tube. I still have that one. I also had a 45 based TSE which made 2WPC, probably the best sounding amp I had, and one that I kept. I had a 300B TSE and an the 845SE, which uses a 45 based TSE for a driver. I still have it, but have not turned it on in 10 years. I had made several SPP based amps, but I have never been thrilled with EL84's for Hifi. Most were sold or turned into guitar amps including a two board monster that made 100 watts. There was also the 300Beast a 25 WPC push pull amp which was a junkyard dog built with junk box parts that defied logic and sounded great. Any attempt to modify it with "good parts" made it sound worse. It ate lightning and died, but I kept it to try to figure out why it sounded so good. The 845SE and the 300Beast were in the 25 WPC class, so that is my target for an SE amp.
My wife considers most of what I listening to as non musical or noise, as did my parents. But then my wife listens to country, and whatever that noise is that she calls pop music.
I have been known to buy music that I don't like enough to listen too regularly, but it makes a good test of an amp / speaker combination's ability to "get right."
Are you using the Toroidy KT88PSE 1.5K opt wired for 3K to do the test? I have both the KT88SE 3K and KT88PSE 1.5K OPTs and have noticed that the 1.5K KT88PSE wired for 3K always does slightly better than the 3K KT88SE. Usually a watt or so more output and slightly lower distortion. My last tests were done with the KT88SE since I do want to build this with both 8 ohm and 4 ohm taps avalable.
In two days of off and on listening to the UNSET board with the Toroidys wired for either 1500 ohms with a 350 volt supply, or wired for 3000 ohms with a 425 volt supply, I think that I like the sound of the 3000 ohm connection better. The difference is not great though. To make matters more confusing I swapped the Toroidys for Hammond 1628SEAs wired for 2500 ohms. As expected these transfomers, which are about twice the weight and size of the Toroidys have better, and cleaner bass, but are slightly subdued in the upper treble. This is not necessarily a bad thing with the "in your face" tweeter in the NS-10M's on some music. The big Hammonds are the lossiest transformers I have ever tested, so even though a 2500 ohm impedance should bring higher power output, the amp is already slightly clipping at 20 watts. If the Hammonds were the same size and weight as the Toroidys I would use them in my amp build since this amp will be used to power the NS-10M's. Unfortunately, now I'm faced with the amp I want in a big heavy box, or a slightly less appealing amp that I could fit into a smaller lighter box.......Maybe I'll make TWO amps.
What cap(s) are you running for C2? I have tried the 120uF electrolytic listed in the BOM, a 270uF electrolytic, 2 100uF motor run caps incombination with those or stand alone. I was targeting at least 300uF for C2 but keep coming back to just the 2 motor run caps only. I would prefer a solution that uses less space though.
I am on the fence regarding if I want to build this as an integrated amp or as a stand alone for use with a pre. The SSE that the UNSET will replace is integrated with a volume pot, source selector to switch between the dac or phono, and sub out. The biggest problem with this setup is if my wife turns it off to watch TV the dac may still be streaming and signal will pass through to the subwoofer which can be very annoying.
Other things I am thinking about is if the uNSET needs a fan and if it should have exposed tubes with those top caps.
I am on the fence regarding if I want to build this as an integrated amp or as a stand alone for use with a pre. The SSE that the UNSET will replace is integrated with a volume pot, source selector to switch between the dac or phono, and sub out. The biggest problem with this setup is if my wife turns it off to watch TV the dac may still be streaming and signal will pass through to the subwoofer which can be very annoying.
Other things I am thinking about is if the uNSET needs a fan and if it should have exposed tubes with those top caps.
I have used multiple different combinations including nothing at all, just a bench power supply wired into the board with C2 removed so I could test at voltage level that would make my current C2 very unhappy. The listening tests over the past few days were done with a 120 uF 500 volt Panasonic electrolytic cap in the board and a 100 uF motor run cap in parallel. For space limited builds I usually use the biggest electrolytic that I can fit into the board and parallel it with the biggest polypropylene cap that I can fit under the deck. In my 45 based TSE that is a 40 uF 400 volt ASC cap that I got surplus. I have some polypropylene "Unlytic" caps that I bought from another used here many years ago that will find their way into my UNSET when I build it.
When I had a similar setup in Florida I simply plugged all of the "HiFi stuff" including the powered sub, into its own power strip which was turned off for TV watching so the TV played through it's own speakers. The setup in my work room used a line stage which could be turned off to silence everything. I used it primarily as a buffer with multiple outputs, each with it's own volume pot so I could feed a common signal to one (or more) of several amps in my setup. I also had a box full of relays and switches to select one or more sets of speakers for each amp.
I have 1 amp pico fuses wired in series with my 1 ohm cathode resistors. Yesterday I shut off the fan that blows air across my board and let it get hot for about an hour with no signal applied (worse case heat). The idle current will creep upward as the fets get hot which lowers their threshold voltage. I was curious if this could lead to a runaway condition. The idle current was set at 110 mA after a 5 minute warm up and it leveled off at 121 and 123 mA after an hour. The heat sinks were too hot to touch with readings between 70C and 84C with my $20 IR thermometer. The same thermometer shower readings of 240 to 260C on the glass of the output tubes near the caps. If real, this is a bit too hot, but I am running them at 40 watts, which crept up to nearly 46 watts over time. I think that some of the heat sink heat was IR absorbed from the nearby tubes. My other unfinished UNSET board will have the heat sinks on the bottom, but I will try using the chassis deck as a heat sink in a build, and add a finned heat sink to the top if needed.
I am not a fan of fans, but it is possible that I use one here for entirely different reasons than hot heat sinks. When I lived in Florida there were three people living in the house, myself, my wife, and my daughter and two cats. All were taught to respect electricity, that it could kill you dead, and cook your body to a crispy state if touched. The door to my workroom was closed whenever I was not there, and there was this big red killswitch button on the corner of the bench which depowered everything in the room except for the lights and computer. The bench was "hot" only when I was there. By high school my daughter was in the marching band and a house full of kids armed with musical weapons was not uncommon. They knew that the workroom was off limits and that stuff in there could kill you, but all of the amps and musical gizmos that they played with came from that room so it was respected. I made guitar amps for the kids that wanted one, which helped in the respect department.
Today I have 4 grandkids. They do NOT know about, fear, or even understand high voltage electricity. This is prevalent in today's "YouTube and TikTok generation" where stupid deadly stunts seem commonplace. People are often seen playing with live electronics, and most of that is relatively harmless 5 volt or 12 volt stuff. We also see videos of people poking around in a live microwave oven, or using a microwave oven transformer to burn wood which recently killed a couple and burned down their house. Unfortunately there is rarely any differentiation between these two extremes given.
Again, my entire workbench is cold whenever I'm not there working. Potentially deadly things like my UNSET test board have its power cord removed when I'm not working on it. I do this out of habit since I got zapped a few times from simply yanking the cord from the power strip. All power cords are black and it's too easy to unplug the wrong one. Anything built for use in my basement workspace must be "grandkid proof" which means insulating them from high voltages and hot tubes. That's the real reason my 845SE amp has remained cold and up on a shelf. It has exposed 1050 volt wiring on both chassis and 845 tubes that get real hot. Now that I have UNSET, I'll probably part it out or sell it for parts.
An UNSET based amp, and a push pull big brother or two will find their way into the rack of equipment that is part of my computer based music workstation. There is already a rack mount PC in that workstation which contains multiple fans. The UNSET that I make for this workstation will be rack mount and probably be built into a rack mount PC cabinet, so the enclosed tubes and top caps are a given. A PC fan or two may be used to direct tube amp heat out the back of the rack so it doesn't cook everything above it in the rack.
If I build a conventional looking tube amp it will have a cage over the tubes. The cage will probably be a wire office basket from Ikea, Walmart, or a thrift store. To cage or not to cage is a decision best made depending on your use case. Kids, cats and anyone not understanding the danger must be trained not to go near an uncaged amp, or protected from touching it. It took two years of blasting kitty in the face with a water gun to train him not to go into the work room. Did that keep him from jumping onto my turntable in the living room when it was in operation? No, it resulted in the destruction of a Garrard Zero-100 in the mid 70's which turned out to be a good thing since I bought a Technics SL-D2 which I still use today.
Last edited:
In my case the UNSET will live in an entertainment center behind some doors along with my dac, streamer, and turntable. The SSE is there now and nothing gets too hot. I do leave the doors open a little and there is a good sized hole in the back for some airflow. Of course the UNSET will put out more heat and have a bit less ventilation with some sort of metal mesh/cage over the whole thing. I built it with everything on top so it will take up less space when covered, essentially the PCB could sit lower in the chassis. I do have the other board with heat sinks on the bottom that I'll probably build more conventionally and it will end up in the office or garage.
I again tried using a 120uF electrolytic cap in combination with a 100uF motor run cap for C2. This was after I did some cleanup on the board by replacing a number of resistors and such that I had paralleled or put in series to get the right values. I did some measurements after and everything was slightly worse. I pulled the 120uF cap and put back in the other motor run cap. So 2 100uF motor run caps for C2. Measurements were back down to where they were before. Not huge changes in the measurements(~0.03 1W, ~0.09 10W, ~0.8 18W) but for whatever reason in listening I also preferred using just the film caps for C2 so I am going to keep it like this.
spiggs,
Interesting that your ears can tell consistently whether you are using film or ‘lytic caps in C2, and you can measure a change in distortion. My still non-operational board has 390uF/500V Kemet electrolytic cap soldered in C2 - perhaps I will dig out a nice 200uF Unlytic cap I have somewhere and compare for my self.
Interesting that your ears can tell consistently whether you are using film or ‘lytic caps in C2, and you can measure a change in distortion. My still non-operational board has 390uF/500V Kemet electrolytic cap soldered in C2 - perhaps I will dig out a nice 200uF Unlytic cap I have somewhere and compare for my self.
As to the measured THD differences, it might just be a reduction in hum or noise at those levels. I get a 0.1 or 0.2% reduction in THD by turning off the Walmart quality LED shop lights and every time I mess with the amp I get a slightly different reading due to the messy breadboard layout picking up crud.
Years ago I could hear the obvious difference between a TSE or an SSE with only the 120 uF electrolytic, or the electrolytic and the motor run cap in parallel. A degenerative middle ear issue called Meniere's Disease has rendered me nearly deaf and with some serious balance issues. I can still hear the difference between obvious things if it's loud enough, but hearing caps is out of the question for me now. I can still hear the difference between an SSE and a TSE-II of similar power levels when played through the Yamaha NS-10M's, and the UNSET is still the punchiest, most dynamic sounding SE amp I have ever heard, but I have not done a A/B comparison with a KT88 based SSE with CFB enabled yet. To be fair the test would have to use the same OPT's for each amp.
I still have one more amp technology I want to conquer before I hang up my amp building hat.
My last frontier in tube audio is a high efficiency amp using a switched power supply rail (class G) or a modulated supply rail (class H) on a class A output stage. This was revealed in a magazine article I wrote 15 years ago. For it to work the output stage must be a cathode follower which requires LOTS of drive voltage. So far, I'm looking at a small sized UNSET for this driver, so the UNSET will be on the test bench for the near future. I will move from big output tubes to small ones when I get caught up on other stuff.
Today is May 1, which starts the countdown to the Dayton Hamfest which is less than 3 weeks away. I plan to be at the swap meet in a small cargo van stuffed to (or maybe a bit over) capacity with stuff that I don't see myself using for the rest of my life. Much of today was spent going through my tube collection and culling the herd a bit. Yes, I will be selling sweep tubes, dampers, and other tubes that I will never use for as little as $1 each. If I can get to them there will be some transformers for sale too. Lots of old radio stuff, some computer stuff, and even some ham radio stuff too.
I remember you talked about it, but unfortunately that page is now down: https://circuitcellar.com/microchip2007/winners/MT2209.html
I will take a look at what has been written here:
https://www.diyaudio.com/community/threads/augmented-cathode-follower.114021/
https://www.diyaudio.com/community/threads/class-g-class-h.116042/
and some more external links:
https://www.labjc.com/wp-content/uploads/2021/02/040CathodeFollower-1.pdf
http://www.classeradio.com/modulate.htm
A SE amp with high efficiency, that would be really great!
I will take a look at what has been written here:
https://www.diyaudio.com/community/threads/augmented-cathode-follower.114021/
https://www.diyaudio.com/community/threads/class-g-class-h.116042/
and some more external links:
https://www.labjc.com/wp-content/uploads/2021/02/040CathodeFollower-1.pdf
http://www.classeradio.com/modulate.htm
A SE amp with high efficiency, that would be really great!
The one difference I did not try to control for is the physical placement of the caps. The electrolytic is of course mounted on the PCB, since these are snap in it was easy to install or remove for testing and they fit real snug since a little solder remains from un-soldering. I tried both a 120uF and 270uF in place. The 2 100uF motor run caps are underneath the PCB on my bench connected of course by wires to the terminals. Perhaps it is just that the top mount cap picks up a bit more crud in my bench top setup or the metal motor run cap cans keep it out better? I tried the test with and without a highpass filter in REW of 200hZ with the same results. I would like to keep the wiring to a minimum so to me off board caps are not ideal beyond the space issues they present but as is they don't seem to be a penalty in THD. Of course REW is the only measuring setup I have so if anything else is at play I have no idea.
As for sound my impression is this. Electrolytic cap alone sounds good, add the film cap and the bass sounds a touch fuller with a little more punch, take away the electrolytic and the vocals sound a touch smoother. This is all subjective and tests are using my cheap 2 way garage speakers so take it as you may. I certainly don't have a golden ear for such things and would be curious if anyone else tries this what their impressions are. Good thing is the motor run caps are cheap, those Unlytic caps mentioned seem to go for a bit more but then again you can get a 290uF 750VDC rated cap for just a bit more than the cost of 2 100uF motor run caps if you buy from one of the surplus places. They are big though.
George, do you think you will offer a revised UNSET PCB with the CCS and feedback mods integrated? That next amp sounds interesting, hope I can follow along.
Don't tell me you will have a box of excess 26HU5s or 26LW6s with you to sell for cheap. California is just too far from Ohio for me to make the trip.
I will eventually make another UNSET based board or board set. In the early discussions about what the users wanted from the UNSET design, most seemed to prefer a modular approach, separate boards for each section, driver, power supply and output. This offers the most flexibility. I just happened to have this board nearly done at the time, and there is still lots to learn with it. Adding the current known mods to the existing board will probably happen, but not until I'm sure that's all we need.
No 26HU5's or "LW6's. I'm clearing out stuff that I will likely never use. So far there is a box of 6GV5's, a box full of 6KY8A's, a box full of 12AV5GA's, and a box full of assorted damper rectifiers. There will be more boxes. I may bring my collection of 4D32 transmitter tubes as they look cool but have not found their way into anything more than a test board in the 15 years that I have owned them. I will keep adding stuff to the hamfest collection until I reach one cubic van load.
Have you seen this link?
https://www.diyaudio.com/community/...-uprocessor-with-tube-amp.379145/post-6840668