Okay, not fair! I have to land on an output tube! I don't need 60w out, maybe 20 or so, and I'm thinking the 300B would do that very well. I would need one more pair of 300b's, but other than that, with AC on the fils, I'd have all the parts. The B+ of 465 with fixed bias might be a bit hot. Having ordered the power transformers, an option would be to stick a tube rectifier like a 5R4 in there to drop below 450, even consider cathode bias I guess to maintain a safe tube operation.
My budget does not support the use of expensive tubes. I have used Sovtek and Shuguang 300B's. Neither like to be run over 400 volts. They will go into a runaway condition. I had about 440 volts, so I designed a mosfet voltage regulator which kept blowing up. I simply added another choke to the power supply and ditched the regulator. The CLC supply became a LCLC supply and the B+ is 375 volts. I used the 6.6K CT OPT that I was originally using with 6L6GC's and KT88's. It worked great even though it was a $16 POS from Ebay. I have tried several other opt's including some UTC LS57's but there is some synergism amongst the junk, so the guitar amp OPT's remain. That was one of my favorite amps, seeing daily use for about 5 years until it blew up.
The 300B's will require more negative bias than the KT88 and a bunch more drive. I have never built this exact circuit, but it should be possible to make it work. I used a combination of a 300B design that Kevin Kennedy (kevinkr on this forum) published in VTV a long time ago and a KT88 design that I had been using for guitar amps back then. The KT88 design only made about 35 watts because the cheap Chinese KT88's of the day would blow up (sometimes quite violently) if pushed beyond 35 watts.
The 300B route will take some tinkering, but it shouldn't be too hard if you have experience with this stuff. The KT88's should just work since it is a proven design.
So I will cloud your mind with yet another possible path. Triode wired KT88's. You might get close to 20 watts in class A with a 6600 ohm OPT, and it will sound real nice. Here you have the option of building the amp as designed and then changing a few wires for triode operation. I have a P-P KT88 amp that is triode wired and I like it too. I had it set up for fully differential operation of all 3 stages resulting in about 15 WPC in triode class A. It sounded real nice but didn't have enough juice for my ineficient speakers. I set the amp up for triode AB2 and got about 50 WPC. That is how it works today, but I have since built some high efficiency speakers, so the fully differential path will be explored again.
Now we're talkin'....that's what I'm planning on building next, something like a power-driven Mono Bill, or Sgregory's Opus with Trioded KT88's, or perhaps a MOSFET followed Piondexter KT88 music machine.
What's the advantage (if any) of a 3-stage Mullard 5-20 design over a 2 stage design? Better linearity/less voltage swing for the driver/LTP sections? If one can live with lower input sensitivity (ie power amp duty), are three stages req'd for trioded KT88's?
3 stages gives better possibility of GNFB. Certainly don't need much of the gain. The 12AT7 in the LTP is limited in the amount of gain with the designed B+ of these amps.
FYI, The OPUS might go triode this winter. I hate to change too much as I have not tired of it yet as it is. I even have discrete CCS and gyrators to go in as loads but can't get myself to make the change from the chips
FYI, The OPUS might go triode this winter. I hate to change too much as I have not tired of it yet as it is. I even have discrete CCS and gyrators to go in as loads but can't get myself to make the change from the chips
Last edited:
My design wasn't exactly inspired by the Mullard circuit. In fact there aren't too many similarities. The design started out as a combination of several works in progress and was most influenced by Kevins 300B design since it was a working amp that matched a design that was already in my head at the time.
I am counting the output stage, so there were two driver stages plus the output stage. Two total stsges falls short of total gain in most cases and squeezing maximum gain out of a two stage amp doesn't lead to best linearity. Two 6SN7's and a pair of KT88's per channel just sounds good. One 6SL7 and one 6SN7 makes a good driver for tubes that need more drive like the 300B. If you like miniature tubes use a 5751 and a 6FQ7.
I had come to the conclusion that cascading two LTP's together just works great and sounds phenomenal. The 300Beast was my first attempt. It was never planned as complete amp, hence the total junk box parts selection. It was built over 10 years ago and worked so good that I still have it. It is the only amp from that era that I still have. I have been experimenting along the same lines ever since. Apparently some other great minds have come to the same opinion. I mentioned Kevins design. Morgan Jones' Crystal Palace amp folows the same concept, as does a design by Allen Wright.
I been working on a replacement for the 300Beast when a request for a similar design was posted here. That led to a year long collaborative effort across two continents where the amp design went from idea to working amp right on these pages. It is here:
http://www.diyaudio.com/forums/tubes-valves/133034-6l6gc-ab2-amp.html?highlight=6L6GC+AB2
I am still experimenting along these lines and I have several variations breadboarded up now.
The main reason for going this route is dynamic range. You can only get so loud. This limit is forced by the power output and speaker efficiencies. In order to expand the dynamic range further, you must be able to get quiet too. This limit is usually set by the hum and noise in the background of all electronic devices. There is a limit to reduction of these effects, but using fully differential circuitry allows some of the effects to cancel. This is especially true of power supply hum and noise.
It is not too difficult to build a fully differential power amp. I had one up and running. A fully differential class A power amp has the best "downward dynamic range" but is limited in power output by the class A output stage. My speakers at the time worked best with a trioded AB2 output stage using KT88's with zero feedback anywhere in the amp. The new speakers may like something else.
It is possible to build a fully differential audio system from the phono carttridge to the output transformer. This has been done in the solid state world, and may have been done in an all tube design, I don't know. I have seen talk of such things but haven't seen evidence of one in the wild.
Allen Wright was a big fan of differential from the cartridge to the output. I have built his PP-1 with great success. I'm currently running a variation on that - same front end but I also have a CCS on the output tubes to bias them (EL34's) at about 60ma each. The amp is dead quiet and has great dynamic range. I have successfully substituted KT66's and KT88's as well since the CCS means no bias adjust needed.
Thanks George, you just answered my next question, which was going to be "what are the advantages/disadvantages of the 5-20 topology compared to cascaded LTP's like Chrish's amp?..........
So, back to the 5-20 3 stage vs LTP as driver 2 stage differences.......You usually need 3 stages to get enough total gain to allow for GNFB. A no feedback triode design can get away with just a LTP stage possibly at the expense of some linearity since we don't burn any gain with feedback. The above assumes that the LTP can provide enough voltage swing and current to keep the outputs happy.
Although the 5-20 topology has 3 stages, it still only uses one coupling cap due to the first/second stage being direct coupled, keeping Nyquist satisfied as far as stability goes, making this implementation straightforward.
Is the above a reasonable summary or did I miss anything?
So, back to the 5-20 3 stage vs LTP as driver 2 stage differences.......You usually need 3 stages to get enough total gain to allow for GNFB. A no feedback triode design can get away with just a LTP stage possibly at the expense of some linearity since we don't burn any gain with feedback. The above assumes that the LTP can provide enough voltage swing and current to keep the outputs happy.
Although the 5-20 topology has 3 stages, it still only uses one coupling cap due to the first/second stage being direct coupled, keeping Nyquist satisfied as far as stability goes, making this implementation straightforward.
Is the above a reasonable summary or did I miss anything?
An LTP has 1/3 to 1/2 the voltage gain as the same tube used in a voltage amplifier circuit. I like LTP's but during the design of the Simple P-P I could not find a reliable way to use a single LTP to drive an EL84 to full power. It is a bit short of gain using a 12AT7 even without feedback. A 12AX7 works but doesn't have enough gain for feedback use.
Pete Milletts amazing red board uses pentodes in LTP but even this can run out of gain with minimal local feedback applied to big output tubes in a 125 WPC build. I have been experimenting with 3 total stages with a pentode as the second stage to get big power (250 to 400 WPC).
Other than the gain in a 2 stage design it is right on. I have made two stage designs work. They used a high mu triode driving a high Gm pentode.
Doesn't using a mosfet follower free up the design to use a low gm, high mu tube for the LTP?.... or do you still need some gm to drive the input capacitance of the mosfet?
Who......MEEEEEEEEEEEEEEEE..... I didn't invent Petes red board....I just took his mild mannered 18 WPC amp and extracted 250 WPC. Then I dialed it back to a more civilized 125 WPC and posted the mods for everyone to copy. Several people have built it with good results.
Since I have a pair of OPT's rated for 400 watts @ 20Hz, I need to find a design that uses them to their fullest potential🙂. I havent decided on exactly what that is just yet. One possibility uses a design that started with the red board, added another stage to boost the gain, and added mosfets to drive the grids out of whatever output tube or tubes I choose to connect. Another possibility uses the previously mentioned 6SL7 - 6SN7 - mosfet follower combination. Either way I want to be able to drive anything through any combination of G1, G2 or both.
I have been using mosfets that are in the 5 to 10 pF range. Most high mu triodes can drive them just fine. I have recently found a mosfet that is under 2 pF and can eat 600 volts. There are 100 of them waiting for my experiments when I get back home. Even a 12AX7 should have no problem with them.
The Simple P-P was to be a sand free design, so it doesn't use mosfets.
This weekend I got to it - cut up both chassis and mounted most large parts on one of them. Still deciding whether or not to paint the chassis (Hammond gray) or the Edcor transformer (awful blue). I probably will, but I'm so close to soldering something, it's hard to resist going full steam ahead! My experience has been that I spend a lot of time trying to make it look pretty and then turn around and hack it up to make something new... I'm already wondering about adding 2x KT88's and a bigger PT. No, that would just be wrong.
I have an early pair of Edcor CXSE25-8-5K OPT's. They aren't the usual boring blue color that all the recent ones are. They are bright metalic blue. It is a color that wasn't too different from a "pimp my ride" Cadillac that used to hang around the neighborhood. Too bad they don't use it any more. The amp would light up the room even before you turned it on!
Reminds me of a Porsche 928-S4 I used to see occasionally. A real "head turner" with electric blue paint and loud exhaust.
jeff
How about going for an Ikea theme and paint the chassis yellow. You can name the amp "elva" which google translate tells me is swedish for eleven. If you are unlikely to have any Swedish friends over, you can throw in a couple umlaut's and put a ring over the 'a'...
😀
(sorry, lurker from the peanut gallery)
Making progress on my KT-88 "Mullard 5-20", this is the bias supply and a couple other PS parts
The chassis has since been painted, as has the blue transformer