I got my TubelabSE pcb finished and powered up last night.
After about three hours, my left channel went dead. I swapped output tubes, and the channel remained dead. Fearing the worst, I measured the adjustable negative bias, and found it to be about (-) 187 volts.
I'd say that 45 was driven pretty well into cutoff.
At the time I made my parts buy, the MOSFET choice was unsettled due to the discontinuation of the 2SK2700. I chose to try the IXYS IXTP1R4N60P that was a suggested choice.
Is this still a good MOSFET choice and did I just get unlucky with a part that succumbed to infant mortality? Or is there a better choice now that some of these boards are seeing some field use?
The board seemed to work fine, this glitch notwithstanding. The only other glitch was that I could not get one of my 5842's to more than about 145 volts. I don't know if that's a tube issue, or another issue to look into - I just set them both to 140 volts and called it good.
I did have a bit of low level hum, but that was not unexpected at this point, given the amp is spread all over my bench, and I am using CRC rather than CLC to cut down on the clip leads during testing. I haven't measured it yet, but you have to put your ear close to the speaker to notice it.
For anyone that is interested, I am using a NOS Stancor 8404 transformer, 260-0-260v at 90 mils, 5v at 2A, 6.3vct at 3A for the power transformer, and I used a 33uf cap at C4. The tubes I tested with are 5AR4 and 45's. This puts my loaded B+ at 308 volts, measured across R4. haven't decided on OPT's yet.
Win W5JAG
After about three hours, my left channel went dead. I swapped output tubes, and the channel remained dead. Fearing the worst, I measured the adjustable negative bias, and found it to be about (-) 187 volts.
I'd say that 45 was driven pretty well into cutoff.
At the time I made my parts buy, the MOSFET choice was unsettled due to the discontinuation of the 2SK2700. I chose to try the IXYS IXTP1R4N60P that was a suggested choice.
Is this still a good MOSFET choice and did I just get unlucky with a part that succumbed to infant mortality? Or is there a better choice now that some of these boards are seeing some field use?
The board seemed to work fine, this glitch notwithstanding. The only other glitch was that I could not get one of my 5842's to more than about 145 volts. I don't know if that's a tube issue, or another issue to look into - I just set them both to 140 volts and called it good.
I did have a bit of low level hum, but that was not unexpected at this point, given the amp is spread all over my bench, and I am using CRC rather than CLC to cut down on the clip leads during testing. I haven't measured it yet, but you have to put your ear close to the speaker to notice it.
For anyone that is interested, I am using a NOS Stancor 8404 transformer, 260-0-260v at 90 mils, 5v at 2A, 6.3vct at 3A for the power transformer, and I used a 33uf cap at C4. The tubes I tested with are 5AR4 and 45's. This puts my loaded B+ at 308 volts, measured across R4. haven't decided on OPT's yet.
Win W5JAG
Others that tried the IXYS IXTP1R4N60P had similar fate. I removed them from my reccomended substitutes list a while back.
The Fairchild part works well at the voltages that you are using, but has been known to oscillate when operated in the 400 volt range. This has zapped a fet while I was messing around with it.
I have 4 TubelabSE amps on my bench currently and I am testing several different mosfets. My current favorite is the Toshiba 2SK3563. It has the flattest "reverse transfer capacitance" VS voltage curve that I have ever seen. It is only good to 500 volts which should be plenty for all possible TubelabSE amps. I have another amp running with IR IRF820A fets in it. It works well also. I have about 5 more types of mosfets to test, which should get done this weekend.
The 5842 tubes seem to vary all ovet the place and I have met a few that wouldn't make more than 150 volts as well. You can adjust the value of the 510 ohm resistor in the cathode circuit to make your tube cooperate.
I have found that some 5842 tubes will oscillate if the input cable is grounded at one end only. The grid stopper resistor value may need to be increased to kill the oscillation. The current value is 1K ohm and I had to go to 5.1 K ohm for complete stability over about 15 different tubes. The resistor should be a carbon composition resistor.
The mosfets have always ran warm to hot on the TubelabSE. There have never been any failures with the Toshiba 2SK2700 parts, but the IXYS parts have failed at high B+ voltages. I have seen the Fairchild parts fail at voltages over 400V as well. To lower the dissipation in the mosfet the 20K source resistor should be increased if running at a B+ voltage over 325volts. I am currently running 30K resistors in an amp at 360 volts and 36K resistors in a 400V amp. With your 308 volts, 20K should work fine.
I will be performing a few more experiments on the TubelabSE amplifiers in the next week or two and I will update the parts list as soon as I have verified everything on several amps.
The Fairchild part works well at the voltages that you are using, but has been known to oscillate when operated in the 400 volt range. This has zapped a fet while I was messing around with it.
I have 4 TubelabSE amps on my bench currently and I am testing several different mosfets. My current favorite is the Toshiba 2SK3563. It has the flattest "reverse transfer capacitance" VS voltage curve that I have ever seen. It is only good to 500 volts which should be plenty for all possible TubelabSE amps. I have another amp running with IR IRF820A fets in it. It works well also. I have about 5 more types of mosfets to test, which should get done this weekend.
The 5842 tubes seem to vary all ovet the place and I have met a few that wouldn't make more than 150 volts as well. You can adjust the value of the 510 ohm resistor in the cathode circuit to make your tube cooperate.
I have found that some 5842 tubes will oscillate if the input cable is grounded at one end only. The grid stopper resistor value may need to be increased to kill the oscillation. The current value is 1K ohm and I had to go to 5.1 K ohm for complete stability over about 15 different tubes. The resistor should be a carbon composition resistor.
The mosfets have always ran warm to hot on the TubelabSE. There have never been any failures with the Toshiba 2SK2700 parts, but the IXYS parts have failed at high B+ voltages. I have seen the Fairchild parts fail at voltages over 400V as well. To lower the dissipation in the mosfet the 20K source resistor should be increased if running at a B+ voltage over 325volts. I am currently running 30K resistors in an amp at 360 volts and 36K resistors in a 400V amp. With your 308 volts, 20K should work fine.
I will be performing a few more experiments on the TubelabSE amplifiers in the next week or two and I will update the parts list as soon as I have verified everything on several amps.
Thanks, George.
The right channel is still soldiering on - I turned it on this morning, and it's apparently played all day without issue.
Is the 2SK3563 available from Digikey or Mouser?
You think the powerdrive circuit could drive an SV811 or SV572? Not that I would try anything like that ...
Win W5JAG
The right channel is still soldiering on - I turned it on this morning, and it's apparently played all day without issue.
Is the 2SK3563 available from Digikey or Mouser?
You think the powerdrive circuit could drive an SV811 or SV572? Not that I would try anything like that ...
Win W5JAG
I got my 2SK3563's from DigiKey.
The PowerDrive version that is on the TubelabSE board should have no problems with the SV811 or the SV572. I have not tried it since I don't have any of those tubes. I have experimented with some old 811A's. They could be made to work in the board, but I never could get them to deliver the sound that I wanted, and the bass was lousy without feedback due to the high plate resistance of those tubes.
The PowerDrive circuit that is on the TubelabSE PC board is limited by the amount of power that can be dissipated in the on board mosfets which limits the quiescent current. As with any follower circuit (tube, mosfet, or BJT) the PowerDrive can source far more current than it can sink. The sourcing ability is limited by the Rds ON of the mosfet and its peak current handling capacity, which even in this case is AMPS! Far more than the grid current rating of any mere mortal audio tube. Any grid current flow through the mosfet and adds to its dissipation, so tubes that require a lot of constant grid current may overheat the mosfet.
The sinking current is limited by the source resistor (20 to 30K) and the negative bias voltage. This is between 5 and 10 milliamps on the TubelabSE board. The sinking current is only needed to discharge the Miller capacitance of the output tube.
I needed a PowerDrive circuit to drive some big transmitting tubes well into A2, so I put the same circuit on a piece of perf board and mounted it to a big heat sink. No problem hitting 100 mA peak grid currents into an 833A.
http://www.tubelab.com/833SE.htm
The PowerDrive version that is on the TubelabSE board should have no problems with the SV811 or the SV572. I have not tried it since I don't have any of those tubes. I have experimented with some old 811A's. They could be made to work in the board, but I never could get them to deliver the sound that I wanted, and the bass was lousy without feedback due to the high plate resistance of those tubes.
The PowerDrive circuit that is on the TubelabSE PC board is limited by the amount of power that can be dissipated in the on board mosfets which limits the quiescent current. As with any follower circuit (tube, mosfet, or BJT) the PowerDrive can source far more current than it can sink. The sourcing ability is limited by the Rds ON of the mosfet and its peak current handling capacity, which even in this case is AMPS! Far more than the grid current rating of any mere mortal audio tube. Any grid current flow through the mosfet and adds to its dissipation, so tubes that require a lot of constant grid current may overheat the mosfet.
The sinking current is limited by the source resistor (20 to 30K) and the negative bias voltage. This is between 5 and 10 milliamps on the TubelabSE board. The sinking current is only needed to discharge the Miller capacitance of the output tube.
I needed a PowerDrive circuit to drive some big transmitting tubes well into A2, so I put the same circuit on a piece of perf board and mounted it to a big heat sink. No problem hitting 100 mA peak grid currents into an 833A.
http://www.tubelab.com/833SE.htm
Thanks, George.
That sounds encouraging and discouraging - I know I have some SV811 but it's the same tube as 811A except for the plate connection and mu, so far as I know, so I guess I will stick with 45's for the immediate future .
I have the good channel wired up to a mono source, got out a better OPT (Hammond 125ESE) and even with a 45 that is barely better than a dud, it sounds really good. I expect I will have zero hum when I put a small choke in the power supply - it's barely audible now with an ear directly against the speaker.
Win W5JAG
One other - I thought it might lack bass response, but I have an AES 11BM8 P-P amp here in the shack, and the TubelabSE with a near dud 45 blows it away top to bottom. The bass response is very good with the 125ESE.
That sounds encouraging and discouraging - I know I have some SV811 but it's the same tube as 811A except for the plate connection and mu, so far as I know, so I guess I will stick with 45's for the immediate future .
I have the good channel wired up to a mono source, got out a better OPT (Hammond 125ESE) and even with a 45 that is barely better than a dud, it sounds really good. I expect I will have zero hum when I put a small choke in the power supply - it's barely audible now with an ear directly against the speaker.
Win W5JAG
One other - I thought it might lack bass response, but I have an AES 11BM8 P-P amp here in the shack, and the TubelabSE with a near dud 45 blows it away top to bottom. The bass response is very good with the 125ESE.
Resistors, caps, semiconductors, and connectors:
Digikey www.digikey.com
Mouser www.mouser.com
Tubes, sockets, transformers and chokes:
Antique Electronics Supply www.tubesandmore.com
The Allied 6K56VG and 6K7VG power transformers, and the cheap Triad chokes:
Allied Electronics www.alliedelec.com
Tubes and sockets:
Esrc www.esrcvacuumtubes.com
I also get some pieces from Ebay like motor run caps, and often find some parts surplus. All Electronics has the screw connectors, heat sinks and often some useful capacitors.
As life goes lately, I never even set foot into my lab this weekend. Sherri's mom has cancer and was hospitalized this weekend, so we spent the weekend getting everything ready for her to travel back up there for another extended stay.
I talked to a "manufacturers representative" that handled the Toshiba line (and others). He told me that several parts were discontinued due to ROHS. The 2SK2700 was one of those. There are still some old parts (containing lead) in circulation particularly in China. The 2SK2700 is still in their catalog, but none of the usual distributors has ever placed an order for them. It is made on their older PiMOS 3 process, which is being phased out in favor of PiMOS 4.
I asked him to find out what Toshiba (or other) mosfet can be used as a substitute. I asked for the best candidates for low "reverse transfer capacitance" at several voltage levels. The 2SK3563 was the best choice at 500 volts, and the 2SK3566 was the best at 900 volts. I have an amp running 2SK3563's but have not tried the 2SK3566 yet. I would believe that the 2SK3563 is the best choice in a TubelabSE because it doesn't go over 500 volts. I still need some higher voltage mosfets for other projects.
I have these and other fets to test in the near future. I will run the usual frequency response, distortion and other torture tests on them in the near future.
You will never know until you try them. The 811A has a Mu of 160 which usually means a high plate resistance. This also causes a high output impedance which some speakers don't like, mine included. The output impedance can be lowered with feedback, which is possible since a high Mu tube has gain to burn.
I spent some time last year trying to make a minimum parts count amplifier using the 6HV5 which has a Mu of 300. The amp worked, but sounded like the boom boom truck down the street when hooked to my speakers.
My TubeLab SE is up and running again, with the Toshiba SK3563 MOSFET's. I have had the parts for awhile, but today was the first afternoon that I could blow off to work in the shack.
The recent acquisition of some 2A3's has made me rethink my build out plans to include the possible use of 2A3's in addition to 45's.
In place of the previous Stancor PC-8404, the amp is now running with a Hamfest special flush mount Utah, part # unknown, 300-0-300, other ratings unknown, but clearly more than adequate from a current standpoint. It has been running for several hours now with 2A3's, and is still pretty much at room temp. I have a small (about 1.5 x 3 inch) RF type heatsink that I drilled and tapped to mount to the filament regulator chip and that seems to be adequate to sink the extra filament current of the 2A3's.
Presently, it is running in an LC, choke input, configuration. The choke is a Hammond 158Q (5H @150 mils). C2 is the specified 150 uF. With a 5AR4 this produces a loaded B+ of 270 volts at about 60 mils per tube. A 10uF cap at C1 bumps B+ up to about 325. With a cap input, I think it could probably run 300B's, but I don't have any. The OPT are Hammond 125ESE.
I don't know if this will be the combination that goes into the final build out, I'm going to play with it for awhile before I start metal work.
Right now, I am just happy to have it running again!
Win W5JAG
The recent acquisition of some 2A3's has made me rethink my build out plans to include the possible use of 2A3's in addition to 45's.
In place of the previous Stancor PC-8404, the amp is now running with a Hamfest special flush mount Utah, part # unknown, 300-0-300, other ratings unknown, but clearly more than adequate from a current standpoint. It has been running for several hours now with 2A3's, and is still pretty much at room temp. I have a small (about 1.5 x 3 inch) RF type heatsink that I drilled and tapped to mount to the filament regulator chip and that seems to be adequate to sink the extra filament current of the 2A3's.
Presently, it is running in an LC, choke input, configuration. The choke is a Hammond 158Q (5H @150 mils). C2 is the specified 150 uF. With a 5AR4 this produces a loaded B+ of 270 volts at about 60 mils per tube. A 10uF cap at C1 bumps B+ up to about 325. With a cap input, I think it could probably run 300B's, but I don't have any. The OPT are Hammond 125ESE.
I don't know if this will be the combination that goes into the final build out, I'm going to play with it for awhile before I start metal work.
Right now, I am just happy to have it running again!
Win W5JAG
To my aging and partially deaf ears 45's sound better than any other output tube that I have used! My only issue with them is that the speakers that I use the most are 87db efficient on a good day. They are near field monitors so 45's work well in a quiet room with most music if you are seated close to the speakers.
I have built at least a dozen Tubelab SE's including a "universal" design that can run 45's 2A3's or 300B's. The 2A3 is my least favorite tube in that amp.
I talked with a user who used Sovtek 2A3's in his Tubelab SE. He had hum problems with them that went away when he switched to 300B's. He was running the Sovteks far above the maximum ratings at 360 to 370 volts. I set my amp up to the same conditions, and put in some Shuguang 2A3's. This was the best sound that I have ever heard from 2A3's. I even tried some vintage RCA (twin plates) at this voltage and they sounded good too. I have purchased his 2A3's to see if they hum here, but have not tested them yet.
I can't recommend running 2A3's at that voltage, certainly not rare or expensive tubes, but I have no problem exploring the upper limits of cheap Chinese tubes. The Shuguang 300B's don't sound half bad either, but they really like 350 volts or so to come alive. I am using a Hammond 272JX which is 300 - 0 - 300 volts, but they are Hammond volts, so on my voltmeter I read 330 - 0 - 330.
Keep an eye out for 307A or VT225 tubes at Hamfests. They can still be found cheap, and they work (and sound) like a 300B when triode wired. it takes a 5 pin socket though.
I found some NX-483's in an old radio at a flea market. It is a 5 volt 45, with 45 quality sound. A 183 is reported to be the same tube, but I haven't found any yet.
I have built at least a dozen Tubelab SE's including a "universal" design that can run 45's 2A3's or 300B's. The 2A3 is my least favorite tube in that amp.
I talked with a user who used Sovtek 2A3's in his Tubelab SE. He had hum problems with them that went away when he switched to 300B's. He was running the Sovteks far above the maximum ratings at 360 to 370 volts. I set my amp up to the same conditions, and put in some Shuguang 2A3's. This was the best sound that I have ever heard from 2A3's. I even tried some vintage RCA (twin plates) at this voltage and they sounded good too. I have purchased his 2A3's to see if they hum here, but have not tested them yet.
I can't recommend running 2A3's at that voltage, certainly not rare or expensive tubes, but I have no problem exploring the upper limits of cheap Chinese tubes. The Shuguang 300B's don't sound half bad either, but they really like 350 volts or so to come alive. I am using a Hammond 272JX which is 300 - 0 - 300 volts, but they are Hammond volts, so on my voltmeter I read 330 - 0 - 330.
Keep an eye out for 307A or VT225 tubes at Hamfests. They can still be found cheap, and they work (and sound) like a 300B when triode wired. it takes a 5 pin socket though.
I found some NX-483's in an old radio at a flea market. It is a 5 volt 45, with 45 quality sound. A 183 is reported to be the same tube, but I haven't found any yet.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.