Hi Francois,
I think George has some other ongoing projects or personal issues, because during the years he ha sproved to give full support.
But yes, I've had similar feelings as you are having. From my side I can only report what I've seen from simulations, because I'm still struggling to finish the new house, and some travels for work are not helping.
but... I've thought on some issues I've seen reported here, and one of them is the blown P-Mosfet on bottom of the output tube when the driver is overloaded (so the Vpp exceeeds 250V). The best solution I've found on the paper (and the bytes of the simulator) is to use a Rod Coleman's shunt cascode driver, because you can get highly linear output with only low order harmonics, because you can limit the highest voltage going to the downstream voltage, and (I need to test its stability) you could also DC-couple the driver and the output tube.
So at least the reliability of the source follower should be solved.
I've also found that some triodes work very well in Rod Coleman's shunt cascode.
The 12AT7 seems to have a strong 2nd harmonic, dominant until full power.
Other triodes seems to be more neutral like triodized pentodes Rod tested and suggested.
I don't want to hijack the thread, just share what I've thought and not tested yet.
I think George has some other ongoing projects or personal issues, because during the years he ha sproved to give full support.
But yes, I've had similar feelings as you are having. From my side I can only report what I've seen from simulations, because I'm still struggling to finish the new house, and some travels for work are not helping.
but... I've thought on some issues I've seen reported here, and one of them is the blown P-Mosfet on bottom of the output tube when the driver is overloaded (so the Vpp exceeeds 250V). The best solution I've found on the paper (and the bytes of the simulator) is to use a Rod Coleman's shunt cascode driver, because you can get highly linear output with only low order harmonics, because you can limit the highest voltage going to the downstream voltage, and (I need to test its stability) you could also DC-couple the driver and the output tube.
So at least the reliability of the source follower should be solved.
I've also found that some triodes work very well in Rod Coleman's shunt cascode.
The 12AT7 seems to have a strong 2nd harmonic, dominant until full power.
Other triodes seems to be more neutral like triodized pentodes Rod tested and suggested.
I don't want to hijack the thread, just share what I've thought and not tested yet.
I am looking forward to seeing more builds and testing of the UNSET design myself. For now I am happy with the performance of my amp. I have spent a lot of time testing various settings and parts combinations with my very limited set of skills and have blown up a number of parts in the process. My desoldering skills have improved immensely. So obviously I fall into the camp of "fools go where angels fear to tread" and it has been a learning experience. As such I will enjoy seeing where this design lands and what more experienced builders find along with more tips and tests from George.
As it stands my UNSET with the 26HU5/12GNA7 combo has played for hundreds of hours, it sounds good and is stable. Aluminum plate has been ordered and I have a basic layout in mind for the chassis. It will be a challenge but I am trying to fit it all within a 12"x12" footprint and build it as an integrated amp with source selection, volume control, and sub out.
As it stands my UNSET with the 26HU5/12GNA7 combo has played for hundreds of hours, it sounds good and is stable. Aluminum plate has been ordered and I have a basic layout in mind for the chassis. It will be a challenge but I am trying to fit it all within a 12"x12" footprint and build it as an integrated amp with source selection, volume control, and sub out.
spigg,
Thank you for your fearless experimentations, measuring the results and sharing what you have tried and learned. You made this thread very enjoyable and educational. I have always looked forward to your posts here and enjoyed reading them. Glad your UNSET with the 26HU5/12GNA7 is stable now and performing satisfactorily. I have a feeling (perhaps unjustified) that the driver section of the UNSET is not quite optimized yet, so, I will need to get the right pieces and learn to use REW so that I also could try things and get objective measurements of the results.
Your 12”x12” UNSET build will be something to behold! Good luck with it and keep us posted.
Thank you for your fearless experimentations, measuring the results and sharing what you have tried and learned. You made this thread very enjoyable and educational. I have always looked forward to your posts here and enjoyed reading them. Glad your UNSET with the 26HU5/12GNA7 is stable now and performing satisfactorily. I have a feeling (perhaps unjustified) that the driver section of the UNSET is not quite optimized yet, so, I will need to get the right pieces and learn to use REW so that I also could try things and get objective measurements of the results.
Your 12”x12” UNSET build will be something to behold! Good luck with it and keep us posted.
@Tubelab_com
I was thinking on how to overcome the core saturation limit shown by Toroidy, and going through a SE transistor amp I've found something that should be helpful.
Can I have your opinion? It should be helpful here too.
https://www.diyaudio.com/community/threads/a-way-to-help-core-saturation-in-se.390419
I was thinking on how to overcome the core saturation limit shown by Toroidy, and going through a SE transistor amp I've found something that should be helpful.
Can I have your opinion? It should be helpful here too.
https://www.diyaudio.com/community/threads/a-way-to-help-core-saturation-in-se.390419
Unfortunately, I am stuck in a situation that doesn't appear to be getting better. On top of "unhappy wife syndrome" my daughter and her family of 6 have become more and more dependent on me for financial and physical assistance. My wife keeps starting projects that she can't finish so that I have to. One is to build a "music lab" here in the basement so that my daughter can teach music to her 4 kids and their friends. That is why resurrecting a 25 year old Roland JV-1000 took precedence over everything. Next up is either fixing up a defunct Simmons digital drum set or reassembling my daughter's conventional drum set from the 90's and outfitting it with mesh heads and MIDI triggers. There is a mixing console and a bunch of guitars too. We have a flooding issue and a defunct mobile home on our property that need attention. The flooding issue is real and getting worse with possible water intrusion into the basement where my lab is, so it looks like I'll be digging a 100+ foot trench across the property to bury a 4 inch drain pipe. Nobody will do this job since it must cross the septic system's drain field, so I need to do it by hand. The mobile home is now a storage space where my tubes and parts are stored. The water leaks need to be fixed. Any dreams I keep having about building a music synthesizer, more and or bigger tube amps, etc have again been squashed for the foreseeable future.
Most of us are doing this as a hobby is true, and so have I for most of my life. I have also had some kind of hobby related business since the late 70's that subsidized the cost of my hobbies. DMA engineering made boards for the SS-50 (SWTPC) computers and several other electronic kits including a DIY satellite TV system. G and S programming made some useful little gizmos that used PIC chips. Both faded with the market for their products. Tubelab got started in the 90's as way for all the technical people at Motorola to build my vacuum tube HiFi and guitar amp designs. Then in the early 2000's one of my friends said, "Why don't you make a web page for that stuff." That evolved into Tubelab Inc. The market for DIY vacuum tube HiFi amps has been declining for the past several years for several reasons. It has also been cyclical, peaking in the winter months, and dependent on currency exchange rates. Tubelab has never made more than $1000 in profit in any year and lost money in some years. That was fine when I had a 6 figure engineering salary.
That ended in 2014 when Motorola was downsizing, and I got the offer that I could not refuse leaving us with no earned income. I now live on Social Security and a small pension. Tubeleb (the company) MUST pay for itself, or it will end. A couple small contract engineering jobs extended Tubelab's life. Sherri, my wife is 50% of Tubelab and has been increasingly in favor of ceasing operations, as it takes up more and more of my time for little return in satisfaction or income. Tubelab has been selling boards for over 15 years. I still answer questions and support boards in amps that were built long ago. After the JV-1000 was done, I put the UNSET back on the bench. That lasted 1 day. It has been replaced with a TSE-II so that I can evaluate alternative regulator chips since the Microchip part in the TSE-II is now unobtainable until June of next year. I must now explore alternatives to the IXYS 10M45 chip (real hard to find) and try a few different mosfets in the TSE-II.
All of my shipping boards were thoroughly tested before being released into the wild. That has made for a smooth transition and relatively easy long term support. This was NOT the case with UNSET. I usually design something, test it thoroughly, make a few prototypes, get some early adopters to build them before they are openly discussed, then if all is well, I will make the board available as a Tubelab product.
The UNSET is a new concept and therefore takes some time to fully understand despite having played with it for over two years. I decided to discuss it here, post the design and the concepts it uses, to see if anyone else wanted to run with it because I was already short of time to work on it. Instead, this brought many requests via the forum, PM and email for UNSET boards. When I made the UNSET beta boards available, I tried to make it clear that support could be sporadic. Supporting the existing boards must take priority. The original TSE and TSE-II could run three different output tubes and one driver tube. I had been building completed amps for a few years with DIY PC boards before ever getting some professionally made. It took some time to know every possible way someone could use that board. Some users found ways to make things with it that I would never have thought of. I even used it as a driver for a bigger SE amp that uses an 845 or 211 tube. I still get requests for a complete build package for that 845 SE amp.
The UNSET board, much like Pete Millett's original Engineer's Amp board can use one of several driver tubes and one of several dozen output tubes. It took me and many other builders about 5 years to learn everything there is to know about Pete's board, and to squeeze up to 250 WPC from his "18 WPC" build. It would take a similar time frame, and considerable money to try every tube and transformer combination that is possible with the UNSET board, especially when only a small number of people are working with it.
Believe me, I would like to spend some more time with my UNSET boards. I still have two more that are partially built. I have some ideas for using the UNSET or some variation thereof in something even BIGGER, but I need to balance my sanity, family harmony, finances and Sherri's ever growing task list. Right now, Tubelab and electronics in general are not getting much time. "Grandkids in the pool every other day" season is over. Winter is coming. Last year we had a long winter season with long term snow or ice on the ground where I need to dig the trench. It rained last night and there are about 2 inches of stinky standing water there now. I will be working out there whenever I can to fix that mess before it freezes over, and maybe out there busting ice after that.
I see the same thing. The issue has to do with the characteristics of the mosfet in the cathode of the output tube. If you look at the curves for the fet you will see a strong temperature dependence. The hotter it gets, the more current it passes with the gate held constant. The only real way to fix this it some sort of temperature compensation circuit or better heat sinking on the fets. Unfortunately, bigger heat sinks collect more heat from the bigger output tubes.
The UNSET board I have with the smaller 6DQ5 tubes and heatsinks on the bottom did well. When I had the maida regulator hooked up bias was very stable.
With the heat sinks on top using 26HU5 tubes and using the adjustment procedure outlined in post #418 the bias between the tubes has remained within 1mA over the course of a couple of months now. I targeted 104mA and have seen it fluctuate between 101mA to 108mA with line voltage and temperature changes but again only 1mA max difference between the tubes which seems ok but I don't know what the standard is.
Do you think tying the heatsinks together would help? How sensitive are the mosfets to being moved off the pcb with extensions to move the heatsinking away from the tubes?
What about using two pmosfet in parallel with a 1 Ohm resistor between each and tube's cathode to balance them?
Each will dissipate half the heat. Input capacitance will double, but seems quite low to cause slewrate issues.
Concerning the topic of the core saturation, I've found @smoking-amp talking about an Audio Precision patent (# 4614914) that effectively removes winding resistance from the OT primary by positive current FDBK. Has anyone tried this method?
In a box now, with E235L outputs and EF184 drivers.
Homegrown choke and OPTs.
In the mess that is my workshop it sounds pretty good: bass and details I didn't hear before.
And that is using a USBC converter cable and my phone as a source.
I'll bring it to my living room later and hook up my regular streamer and speakers.
Didn't measure anything yet except for bias 😅
Homegrown choke and OPTs.
In the mess that is my workshop it sounds pretty good: bass and details I didn't hear before.
And that is using a USBC converter cable and my phone as a source.
I'll bring it to my living room later and hook up my regular streamer and speakers.
Didn't measure anything yet except for bias 😅
Attachments
Please tell us about your operating conditions when you have time. Good luck.
This just reminded me that I have a box full of E130L tubes. I'm not exactly where they are though.
Hi George, you were chatting above about the difficulties you have justifying the time and energy you put into Tubelab, due to the low returns.
I have just re-read the TV tubes sticky thread on the main site, and many of your posts are like vignettes, that summarise a topic, usually with an anecdote or relatable experience.
With a bit of crowd funding to pay for an editor, couldn't there be a quick win there, gathering your comments, compiling 'Tubelab: ruminations' into a book? You turned it up to 11 so we didn't have to. It would be a refreshing alternative to the stock-in-trade tube books.
I'd invest.
I have just re-read the TV tubes sticky thread on the main site, and many of your posts are like vignettes, that summarise a topic, usually with an anecdote or relatable experience.
With a bit of crowd funding to pay for an editor, couldn't there be a quick win there, gathering your comments, compiling 'Tubelab: ruminations' into a book? You turned it up to 11 so we didn't have to. It would be a refreshing alternative to the stock-in-trade tube books.
I'd invest.
I'm planning to use GU50, because from simulations I can get approximaterly 22-23 Wrms at 450V and 3k Ra.
I have many of them and I've seen many project based on them for their sound and reliability.
They are very sensitive to g3 voltage, that can help having knees of the curves moved up-left, so a better pentode, so a better triode here. Best curves are usually with g3 around 15 V above the cathode.
I'm thinking to use a ccs set at 1mA supplied by 450V and connected to g3, then a 15kOhm resistor to the cathode. What do you think?
I have many of them and I've seen many project based on them for their sound and reliability.
They are very sensitive to g3 voltage, that can help having knees of the curves moved up-left, so a better pentode, so a better triode here. Best curves are usually with g3 around 15 V above the cathode.
I'm thinking to use a ccs set at 1mA supplied by 450V and connected to g3, then a 15kOhm resistor to the cathode. What do you think?
Lets not forget G.A has a style to put technical details into perspective, which ordinarily a non-technical person would find very hard to understand. I owe my entire tube DIY effort to G.As posts and comments. He's got a snarky sense of humor too 🙂
Just checking in. For the first time in something over a month I had a few minutes to listen to UnSet yesterday. Good sound! Good amp !
That's it. No additions to technical observations.
There are probes clipped on in a few places ready to take banana plug terminated leads for testing but once the music starts I just don't feel like measuring anything. Can't even be bothered to try different tubes. What's with that?!
Am using it from 750Hz down. Good all the way. (Yup, 'like the bass too.)
That's it. No additions to technical observations.
There are probes clipped on in a few places ready to take banana plug terminated leads for testing but once the music starts I just don't feel like measuring anything. Can't even be bothered to try different tubes. What's with that?!
Am using it from 750Hz down. Good all the way. (Yup, 'like the bass too.)
That's why I tend to leave mine alone. It just sounds nice, so I don't mess with it, because when I do, I'm likely to break something and I currently only have one UNSET that works.
I have fought ADD / ADHD my entire life. Unfortunately, it has gotten much worse since I retired and lost my daily routine. I don't think I could ever organize the ten thousand or so random thoughts that rattle around inside my head into anything resembling a coherent book. I did manage a magazine article once back when I was working. I took some technology I was working on for use in cell site transmitters and stuffed it into a class A SE tube amp and won a Microchip design contest, back in 2007. It won a prize which led to me writing a magazine article.
I keep thinking about revisiting that amp with some of today's more powerful computing devices like a Teensy board. It should make a 30+ watt SE amp affordable and far less of a room heater than even an UNSET. The article did not generate any interest at all. Either the tech was a "bridge too far" in the sacred world of vacuum tube audio, much like the TSE's mosfet driver was, or my explanation was not clear enough. 15 years have passed, maybe it's time for a SE tube amp with a CPU in it to get over 50% efficiency in the output stage, before the "green police" come after our current 5% efficient SSE amps?
Hi George, there is nothing wrong with having somebody else curate those thoughts in an editor role. It is no different to the role of a Technical Author. I'm sure once something is committed to a document it will start to have its own life.
Does anyone use the rectifier tube on the UNSET Beta board?
I MAY do a revision on the existing UNSET Beta board, then get a few more. Then again, this may NOT happen.
I looked into incorporating the mods that some of us are using in our current UNSET board (CCS load on input tube, Current limiting resistor on screen supply, add fuse on the output tube, and restructure the power supply configuration for power up stability) into the current board. The additional parts will not fit the current board size, which is limited by the mailer envelopes that I use. I'm already at the maximum size, so for the changes to fit something must go away and a totally new board layout is needed. Eliminating the 5AR4 sounds like the easiest way to free up some room on the board while removing over 10 watts worth of heat.
Earlier plans included creating a modular UNSET board set, but that is currently unlikely given that Tubelab has lost considerable money for 2022 and continued operations are not guaranteed, so that new products may be unlikely. The global economy downturn and parts shortages have killed the board sales needed to fund new stuff. Current work involves finding and testing alternative parts for existing boards only to find that they are no longer available.
I MAY do a revision on the existing UNSET Beta board, then get a few more. Then again, this may NOT happen.
I looked into incorporating the mods that some of us are using in our current UNSET board (CCS load on input tube, Current limiting resistor on screen supply, add fuse on the output tube, and restructure the power supply configuration for power up stability) into the current board. The additional parts will not fit the current board size, which is limited by the mailer envelopes that I use. I'm already at the maximum size, so for the changes to fit something must go away and a totally new board layout is needed. Eliminating the 5AR4 sounds like the easiest way to free up some room on the board while removing over 10 watts worth of heat.
Earlier plans included creating a modular UNSET board set, but that is currently unlikely given that Tubelab has lost considerable money for 2022 and continued operations are not guaranteed, so that new products may be unlikely. The global economy downturn and parts shortages have killed the board sales needed to fund new stuff. Current work involves finding and testing alternative parts for existing boards only to find that they are no longer available.
So sorry to hear. I am slightly older than you and I have never seen anything like what is now happening in the world.