Originally Posted by TimJo /img/forum/go_quote.gif So the positive feedback is coming from the cathode current of the phase splitter in that design, correct?
I had never looked at it that way. Thats too cool.
Last few years, the problem with this kind of collaborative effort is that when you reach the proof of concept stage, digi01 starts selling boards and you get bogged down supporting builders of an unfinished project.
Originally Posted by burgunder /img/forum/go_quote.gif I think it would be great if you and the community would design an OPT amp, but I will suggest another desing than parafeed as it seems there might be a kit avaiable with the L'espressivo design at least the amp has had it's own section at DIYforums for some time. But this is only speculation, so a parafeed could still be OK.
It's not speculation. There are two prototypes operating as we speak and another 6 on the way. Things are progressing nicely, but I'll defer to Dsavitsk for any other details, if he wishes.
Originally Posted by dBs /img/forum/go_quote.gif At this point I would like it to be a single ended push-pull (SEPP) topology.
Just a quick thought; there has been some confusion in the past about what SEPP means. Most understand it to mean one triode on top of another, however some have bundled the mu-follower under the SEPP banner. You might already have considered the mu-follower, but if not then consider adding it to your list of possibilities.
Mu follower
6SN7
About 10 or 12mA
Parallel-feed
Hammond 119DA
No 'driver' required for an overall gain about 1.5-ish
Plenty for modern high-efficiency 'phones when input is from a normal digital (2V) source
Originally Posted by ericj /img/forum/go_quote.gif Last few years, the problem with this kind of collaborative effort is that when you reach the proof of concept stage, digi01 starts selling boards and you get bogged down supporting builders of an unfinished project.
When is the last time this actually happened? I thought things with Digi had been dealt with to a certain extent? Feel free to PM me with details but if any member is abusing the DIY forum to the detriment of the general membership I'd like to hear about it. It's not something I'm interested in tolerating.
Originally Posted by tomb /img/forum/go_quote.gif It's not speculation. There are two prototypes operating as we speak and another 6 on the way. Things are progressing nicely, but I'll defer to Dsavitsk for any other details, if he wishes.
For now, beyond the fact that it is a low cost, high performance, tube parafeed amp, details are under wraps -- but we are working hard to get this done sooner than later.
Well, I did some research today into parafeed (//feed) OPT design and I like the idea a lot. A very elegant solution to what seems to have been a long plaguing problem. It seems like a reasonable prospect to me since the series cap can be of a very small value which means either high quality or cost compensation for the OPT and choke. The OPT could be cheaper as well since it doesn't need to be as big and doesn't require an air gap.
I really like the //feed idea. It seems that that so far is favored over the P-P design. I certainly think it would be an easier topology to design, especially for the PSU.
I like this, I feel like we're starting to get somewhere now =D I would hate to bite on dsavitsk's feet though =X
Originally Posted by dBs /img/forum/go_quote.gif I like this, I feel like we're starting to get somewhere now =D I would hate to bite on dsavitsk's feet though =X
You can take a different approach, his design is a single gain stage. Look at the 6EW7 datasheet, it would be ideal as one section is a gain stage and the second a drive. So with 2 tubes stereo you get higher gain for high imp phones and lower output impedance for low impedance phones. Bottlehead's amp is similiar but he uses a slightly different tube(same family) which presents some compromises for headphones.
I will look into that combination, thank you for the suggestion.
You guys are getting me all excited to start XD But there's a lot to do yet; math, CAD, buying an oscilloscope and signal generator (not looking forward to that =P).
Researched Mu Followers a bit today. They sound like their advantages are great and one of the few disadvantages they have would be alleviated by the inclusion of a //feed.
I am going to do a bit more research on the mu though as most of what I found today was mathematical design principles and less of the subjective information (I view both as important as I do believe that the ear cannot be summarized by an oscilloscope).
I am attracted to the fact that a twin triode could be utilized as well to make the design a bit simpler (but only a bit since the part count for a Mu goes up significantly).
Something I have been wondering about and have tried to research but found no information on is if twin triodes that are not strapped have the potential to cross talk between each half? That would be an unwelcome occurrence.
I think Leny might be on to a good combination here.
Originally Posted by dBs /img/forum/go_quote.gif Researched Mu Followers a bit today. They sound like their advantages are great and one of the few disadvantages they have would be alleviated by the inclusion of a //feed.
I am going to do a bit more research on the mu though as most of what I found today was mathematical design principles and less of the subjective information (I view both as important as I do believe that the ear cannot be summarized by an oscilloscope).
I am attracted to the fact that a twin triode could be utilized as well to make the design a bit simpler (but only a bit since the part count for a Mu goes up significantly).
Something I have been wondering about and have tried to research but found no information on is if twin triodes that are not strapped have the potential to cross talk between each half? That would be an unwelcome occurrence.
I think Leny might be on to a good combination here.
Read Broskies, SRPP Deconstructed article. They are very load dependant as the load is in the circuit, I've never had much luck with them. Most folks have switched to CCDA is which is also on Broskies site.
Originally Posted by regal /img/forum/go_quote.gif Read Broskies, SRPP Deconstructed article. They are very load dependant as the load is in the circuit, I've never had much luck with them. Most folks have switched to CCDA is which is also on Broskies site.
Yea, I did read that they were load dependent. If you drop the load too much you enter distortion. They also are very prone to quick cut off distortion that escalates very quickly. With a transformer though, I think there is more control over the load. If there are concerns that the load may drop too low, then I can try to design it with a different turns ratio or a tapped transformer.
It seems like the kind of challenge I am drawn to though XD If you tell me something isn't possible or is very difficult, I am more apt to try, lol.
Ultimately, it is a current source strapped to a basic class A triode if I'm understanding it right. If the source of the problem you describe is the fact that it's active load due to the current source aspect then wouldn't you find the same problem if you did a PRIMM based current source? I haven't heard of this problem very much when those are implemented.
Originally Posted by dBs /img/forum/go_quote.gif I am going to do a bit more research on the mu though as most of what I found today was mathematical design principles and less of the subjective information (I view both as important as I do believe that the ear cannot be summarized by an oscilloscope).
Yup, that age old problem… the ear is not an oscilloscope, and at least 50% of subjective opinion is based on false deduction (eg. “I heard a system containing a 12AX7 and the sound was lush. Therefore all 12AX7s sound lush. Get a 12AX7 because they are a lush sounding tube”).
Quote:
Originally Posted by dBs /img/forum/go_quote.gif Something I have been wondering about and have tried to research but found no information on is if twin triodes that are not strapped have the potential to cross talk between each half? That would be an unwelcome occurrence.
I think it would be difficult for an electron from the cathode of triode A to meander over to the anode of triode B. Having said that it seems some triodes did introduce a central shield, so maybe there were some issues? Maybe capacitive coupling at radio frequency (just guessing)? The later version of the 6SN7 was the 9-pin miniature 6CG7, and that certainly had an internal shield connected to pin 9.
Quote:
Originally Posted by dBs /img/forum/go_quote.gif I am going to do a bit more research…
Just for info: with the mu-follower the upper tube’s cathode is at a high voltage and that might exceed the heater-cathode voltage rating of the tube. In which case designers add a positive off-set to the heater voltage to reduce the difference. That seems to suggest that the lower triodes share one bottle while the upper triodes are in the second bottle.
As you seem set to spend a few $, consider spending a few more on TubeCAD from Glass-Ware (John Broskie).
Originally Posted by dBs /img/forum/go_quote.gif
Ultimately, it is a current source strapped to a basic class A triode if I'm understanding it right. If the source of the problem you describe is the fact that it's active load due to the current source aspect then wouldn't you find the same problem if you did a PRIMM based current source? I haven't heard of this problem very much when those are implemented.
No it isn't analogous to a CCS loaded anode follower, the SRPP is a push-pull design where the load (transformer/headphone/top tube) is seen by the bottom triode. Read the Broskie article it is very enlightening, he explains it much better than I can. You are correct that the best way to deal with SRPP is with an ouput transformer so the ideal load is the same / somewhat independant of the headphone impedance. This has rarely been done because you need a gain stage in front of it, most designs just throw a cap on the end and make it an OTL. When I had an SRPP DAC I did a lot of experiementing and I saw a huge reduction in distortion when I lowered my amp's input impedance to the ideal load, however it still didn't sound as good as a SE stage.
Originally Posted by Leny /img/forum/go_quote.gif As you seem set to spend a few $, consider spending a few more on TubeCAD from Glass-Ware (John Broskie).
Right now I am holding off on making any considerations for the order or priority of any expenses. I will definitely have to look into these considerations and being as I am funding it all myself and I'm only just graduated, it may be a slow process XD I have Electronics Workbench software already for the and I will see how good the functionality is for tubes in there. I suspect it wont be very good though, at which point, depending on cost, it would make sense to look into TubeCAD.
Quote:
Originally Posted by regal /img/forum/go_quote.gif No it isn't analogous to a CCS loaded anode follower, the SRPP is a push-pull design where the load (transformer/headphone/top tube) is seen by the bottom triode. Read the Broskie article it is very enlightening, he explains it much better than I can. You are correct that the best way to deal with SRPP is with an ouput transformer so the ideal load is the same / somewhat independant of the headphone impedance. This has rarely been done because you need a gain stage in front of it, most designs just throw a cap on the end and make it an OTL. When I had an SRPP DAC I did a lot of experiementing and I saw a huge reduction in distortion when I lowered my amp's input impedance to the ideal load, however it still didn't sound as good as a SE stage.
I read the TubeCAD information on the Mu Follower today. The source I looked at yesterday was Morgan Jones's Valve Amplifier book. It looks like he came to the current source conclusion as he neglected to account for the loads influence on the circuit.
Unfortunately, there isn't anything new in the world of tube design but I am hoping to avoid the standard SET but just about everything you read now sources you to the sink known as SET XD.
Ultimately, while TubeCAD did show that the Mu is merely a smidgen variation of the SRPP. He didn't really hit on too many major issues with it though beyond the strict shared resistor value. That being the case, I still feel it's worth considering. //feed would be a fitting complement since each of their respective weaknesses are addressed by the other without undue complexity. At the very least the PSRR would be very low, lol.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
