The "ultrapath" connection refers to a capacitor whose negative terminal is connected to the cathode of the output tube. This is used in a cathode biased amplifier to reduce the effects of the cathode bypass cap.
The TubelabSE has the cathode (filament) directly grounded since it uses fixed bias. This eliminates the cathode resistor, and its associated bypass cap, so there is no ned for an "ultrapath" cap.
I do find that placing a high quality external capacitor electrically in parallel with the last electrolytic in the power supply benneficial. Most electrolytic caps (even good ones) have some internal inductance that can come into play in the upper audio range. An external bypass cap can reduce these effects.
The assembly manual can be considered a work in progress. As I get feedback from users, I add, or change the manual. There are now about 70 boards out in the field, many of which have been built up. I have received several suggestions for improvements, which are (or will be) added. I have been building tube amps for 30+ years, so it is hard for me to anticipate the details that an inexperienced builder will need.
Last weekend I added a page of amps built by customers. So far only one person has sent me pictures. More will be added as they come in.
The TubelabSE has the cathode (filament) directly grounded since it uses fixed bias. This eliminates the cathode resistor, and its associated bypass cap, so there is no ned for an "ultrapath" cap.
I do find that placing a high quality external capacitor electrically in parallel with the last electrolytic in the power supply benneficial. Most electrolytic caps (even good ones) have some internal inductance that can come into play in the upper audio range. An external bypass cap can reduce these effects.
The assembly manual can be considered a work in progress. As I get feedback from users, I add, or change the manual. There are now about 70 boards out in the field, many of which have been built up. I have received several suggestions for improvements, which are (or will be) added. I have been building tube amps for 30+ years, so it is hard for me to anticipate the details that an inexperienced builder will need.
Last weekend I added a page of amps built by customers. So far only one person has sent me pictures. More will be added as they come in.
The 2SK2700 mosfets have gone extinct. A few users have found them at suppliers that specialize in obsolete parts, but I am sure that they cost too much (see below). There are several possible alternatives. I have not tested each one yet, so I am relying on info supplied to me by users. I have purchased some of each of these, and have a test amp set up on my workbench. If I don't have to work this weekend, I will test them and put the results on the web site. If I am at work this weekend, I plan to test one fet at a time over the next week. I will be away for 2 weeks in mid August, so it will be done before I go.
Fairchild FQP1N50: This is a great HV mosfet. I have personally used these in a TubelabSE and they worked great. Another builder reported RF oscillation when using this fet, which was cured by adding a 1K ohm resistor in series with the gate lead. This can be done two ways, you can cut the trace on the PC board and add the resistor on the bottom of the board, or you can cut the gate pin short on the FET and add a 1/8 watt resistor in place of the gate lead.
Fairchild FQPF2N90: These are supposed to work, but I have not tried them yet.
IXYS IXTP1R4N60P: These are also supposed to work, but I have not tried them yet. Hard to find outside the US.
Infineon SPP02N60S5IN: Again users report success with these, but I have not tried them yet.
Supertex DN2540N5: A user in Europe wrote me to say that he successfully used this FET in place of the 2SK2700, AND the IXYS 10M45 CCS IC. This sounded somewhat strange, so I tried it and it does actually work. No modifications are needed except the 330 ohm current sense resistor had to be lowered to around 100 ohms (I currently have a trimpot sky wired into a board). I have not had time for critical listening yet, but it will happen soon.
Other users have had success with these fets, but I would be more comfortable recommending a fet after I have personally tested all of these (and a few more that I bought).
I got an email that contained this source for the 2SK2700's. I have not used this supplier before:
"hongkongsuperseller has a bunch of them, but they're not on his ebay site. I'd gotten obsolete chips from him before, so I wrote this address: hongkongsuperseller@yahoo.com.cn
This was the reply:
From: harry hu hongkongsuperseller@yahoo.com.cn
hi,there
10 X 2SK2700 = $24.99 (shipping & handling included)
2 X 2SK2700 = $9.99 (shipping & handling included)
have a good day
hongkongsuperseller company
harry
I waited until I had them in hand as not to spread rumors of something not really out there, or whatever...
-- they look to be pull-outs, there is evidence of solder on the legs."
Fairchild FQP1N50: This is a great HV mosfet. I have personally used these in a TubelabSE and they worked great. Another builder reported RF oscillation when using this fet, which was cured by adding a 1K ohm resistor in series with the gate lead. This can be done two ways, you can cut the trace on the PC board and add the resistor on the bottom of the board, or you can cut the gate pin short on the FET and add a 1/8 watt resistor in place of the gate lead.
Fairchild FQPF2N90: These are supposed to work, but I have not tried them yet.
IXYS IXTP1R4N60P: These are also supposed to work, but I have not tried them yet. Hard to find outside the US.
Infineon SPP02N60S5IN: Again users report success with these, but I have not tried them yet.
Supertex DN2540N5: A user in Europe wrote me to say that he successfully used this FET in place of the 2SK2700, AND the IXYS 10M45 CCS IC. This sounded somewhat strange, so I tried it and it does actually work. No modifications are needed except the 330 ohm current sense resistor had to be lowered to around 100 ohms (I currently have a trimpot sky wired into a board). I have not had time for critical listening yet, but it will happen soon.
Other users have had success with these fets, but I would be more comfortable recommending a fet after I have personally tested all of these (and a few more that I bought).
I got an email that contained this source for the 2SK2700's. I have not used this supplier before:
"hongkongsuperseller has a bunch of them, but they're not on his ebay site. I'd gotten obsolete chips from him before, so I wrote this address: hongkongsuperseller@yahoo.com.cn
This was the reply:
From: harry hu hongkongsuperseller@yahoo.com.cn
hi,there
10 X 2SK2700 = $24.99 (shipping & handling included)
2 X 2SK2700 = $9.99 (shipping & handling included)
have a good day
hongkongsuperseller company
harry
I waited until I had them in hand as not to spread rumors of something not really out there, or whatever...
-- they look to be pull-outs, there is evidence of solder on the legs."
There is a small fet from ST that also appears to have gate charge specs similar to that of the 2SK2700. I started nosing around myself as soon as I noticed that the 2700s were no longer available from Digi-Key. The part number escapes me, but it is the smallest die in the ST Mesh Fet series. The only disadvantage appears to be that an insulated TO-220 package is not available for that part - too bad, as it makes insulating the heat sink harder. I dislike operating parts with hot heatsinks at tube-type potentials. I'd also considered using the Supertex depletion mode FETs, as I have a stash. It's nice to know somebody has succesfully tried them. I'm not building a TubelabSE, but I have an app that benefits from a similar buffer scheme.
The ST FET number is STP2NK60Z. Both Mouser and Digi-Key have them. The capacitance specs are very similar to that of the FQP1N50 (just a touch higher). Big differences are a 600V rating and integral gate protection zener. The zener feature makes me a bit suspicious, as I have heard of squirrely behavior from integral protection zeners when they are actually used for their intended purpose. This was with the VN10KM, the old Siliconix part. Maybe ST has a better implementation. I'm placing a big parts buy soon so I can complete some projects, so I'll probably pick up both the Faichild and ST parts and see how well they work.
)
Yesterday I took my Lexan TubelabSE apart and started testing mosfets in it. I was on my fourth set of fets when I accidentally let the magic smoke out of the amp. There are two resistors blown in half, and the definite smell of toasted silicon. I won't have time to fix it until I get back from an upcoming trip.
Before I blew it up, I ran the Toshiba 2SK2700, the Fairchild FQP1N50, the IXYS IXTP1R4N60P, and the Supertex DN2540N5. I listened to each fet with 45's and 300B's. Music varied from mellow to metal. I pounded it into clipping (where oscillation would show up if it was going to). I probed it with the scope (that's how I blew it up) and voltmeter.
I can say that all four fets worked great with no trace of oscillation. The bias adjustment was identical with the Toshiba 2SK2700, the Fairchild FQP1N50, and the IXYS IXTP1R4N60P. The Supertex part required bias adjustment since it is a depletion mode fet.
I could not hear any obvious differences between the 4 different fets, but the turnaround time was almost an hour, so A-B testing was out of the question.
I think that any of these parts would work in a TubelabSE.
Before I blew it up, I ran the Toshiba 2SK2700, the Fairchild FQP1N50, the IXYS IXTP1R4N60P, and the Supertex DN2540N5. I listened to each fet with 45's and 300B's. Music varied from mellow to metal. I pounded it into clipping (where oscillation would show up if it was going to). I probed it with the scope (that's how I blew it up) and voltmeter.
I can say that all four fets worked great with no trace of oscillation. The bias adjustment was identical with the Toshiba 2SK2700, the Fairchild FQP1N50, and the IXYS IXTP1R4N60P. The Supertex part required bias adjustment since it is a depletion mode fet.
I could not hear any obvious differences between the 4 different fets, but the turnaround time was almost an hour, so A-B testing was out of the question.
I think that any of these parts would work in a TubelabSE.
RockysDad said:
I have a low power SimpleSE spread across my bench right now.
JJ 6V6S's, 5K load with Hammond 125ESE trannies, 280 volts B+, 470 ohm bias resistors, triode connected, no cathode feedback, NOS JAN GE 6201 randomly selected, and all I can say is WOW! This thing rocks! Dissipation is right at the JJ triode spec of 10 watts.
This was supposed to just be a test config to make sure I had it working before I moved up to higher power, but it sure sounds good just like this. At 240 volts, the bass was thin, but at 280 it really comes alive. I've been listening to rock all afternoon - not really the domain of low power triodes, but I just haven't been able to stop listening to it - it sounds that good.
I had planned on 6550's at very high B+ with my Bendix tranny/choke set, and in the back of my mind was thinking about making some adapters for 6384's and a 6106 for a (nearly) all Bendix amp, but I think there is a lot of merit in building one of these out as a 6V6 amp.
I've read a lot about how good the 6V6 sounds as a triode, and I have to say after listening to George's amp this afternoon, I'm a believer.
I put together a temporary 6V6 test amp several months ago using the cheap 5K Edcor OPT's and the Allied 6K56VG power transformer to verify that the combination works. It is ugly, but sounds so nice that I haven't taken it apart yet. My B+ voltage runs from 310 to 325 depending on the mood of FPL (the power company) at any given moment. I never got the time to make any measurements, but I would guess about 1.5 to 2 watts. Sooner or later I will make a real project out of it and put it on the web site. I am currently using a pair of Zenith 6V6G's that I got out of an old radio, but I think that the old smoked glass RCA's sounded better.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.