With the triode/Jfet input solution in mind, I'm going to spend some time tracing load lines to make sure that the Jfets will survive even under extreme overload/overdrive (happens all the time...) I may need to add clamps on the Jfets to protect them. The Jfet/triode combination is not the most cascodey of cascode solutions, as the Jfet drains will wander around some with drive extremes. but I'm more interested in protecting the Jfets than I am providing an ideal cascode environment. If I use Jfets like the PN4393 and J309/310 with low miller capacitance, this will not be a big concern. With the 2SK170s, I'd have to look into it a little more.
I've been wobbling back and forth as to what output iron to use, then an opportunity came by that settled things. An Ebay seller offered output transformers meant for a 160-200W guitar amp. I nabbed a pair at a price lower than a comparable Edcor offering. Initial measurements indicate that they are 7k:8 units, with a primary inductance of a whopping 600 H, measured with a sine wave input of 100V, 300 Hz. They weigh in at 11 lb apiece. I think I will have low end extension if I want it. I suspect these were intended for use with selected tubes at 700V B+ or so to get the stated output rating. I'll need around 650V B+ to reach my stated goal of 100W per channel. With this turns ratio, any sweep tube worth its salt will laugh at the voltage and current.
Last edited:
Woooooooo - the thread that rises from the dead, looking for brains to eat...
I am now the proud owner of a pair of Edcor CXPP100-8-5k output transformers - big and blue. This may move "Killer" closer to realization. I also have an Antek toroidal monster that would give me around 500V plate voltage. If I use that transformer, I would need a small second transformer for running the source follower output drivers as well as a source for powering four 35V filaments.
I was thinking of using a differential folded cascade stage using either pentode or mosfet inputs as the input amp/driver stage, followed by source followers to drive the screen grids. This would give me DC coupling all the way to the final stage (except for an input coupling cap). I'll simulate using mosfets to see how much drive signal I can wring out of a simple setup like that. I intend to limit the Vcc for the source followers to keep from blowing the screen grids in a "full George" overdrive situation.
New thought - it just occurred to me that you could run the 35LR6 filaments in series using a small isolation transformer, maybe with a little boost assist from an extra winding on one of the other hunks of iron.
I am now the proud owner of a pair of Edcor CXPP100-8-5k output transformers - big and blue. This may move "Killer" closer to realization. I also have an Antek toroidal monster that would give me around 500V plate voltage. If I use that transformer, I would need a small second transformer for running the source follower output drivers as well as a source for powering four 35V filaments.
I was thinking of using a differential folded cascade stage using either pentode or mosfet inputs as the input amp/driver stage, followed by source followers to drive the screen grids. This would give me DC coupling all the way to the final stage (except for an input coupling cap). I'll simulate using mosfets to see how much drive signal I can wring out of a simple setup like that. I intend to limit the Vcc for the source followers to keep from blowing the screen grids in a "full George" overdrive situation.
New thought - it just occurred to me that you could run the 35LR6 filaments in series using a small isolation transformer, maybe with a little boost assist from an extra winding on one of the other hunks of iron.
Last edited:
Look carefully inside your 35LR6's. My Sylvanias have an undocumented feature. I also found some RCA's with the same feature, and some RCA's without it. The heater is center tapped and the CT is brought out to an "internal connection" pin (pin 7?). This allows heater operation from 17.5 volts. I did a little creative surgery on one of Pete Millett's big red board to allow operation of the driver tube heaters from the 6.3 volt winding on an Antek toroid, while the 35LR6's ran from the 6.3 VAC + 12 VDC from a 12 volt switcher wired in series with the Antek's 6.3 volts.
In that board I squeezed 250 WPC from each pair of 35LR6's on 650 volts with a 2500 ohm load. At this power level there was a pale plate glow in a dark room. Not bad since I figure that each tube is eating about 60 watts!
All my amp building is currently on hold since I packed up Tubelab and moved 1200 miles. I won't have a lab for 6 months to a year. Before packing up everything I completed a single channel prototype of my "Killer". It started out as one of many LT Spice simulations, and wound up a working board. It uses a pentode LTP input/driver stage and sweep tube output stage. Both G1 and G2 are driven, solving the blown screen grid problem. Output tube plate to driver plate feedback is used to reduce output impedance.
The simulation and prototype used 6EJ7's and 13GB5's because they are plentiful, cheap, and the models are very good. I may choose different tubes for the final build. Does it work? Well the lowly $1 13GB5 can crank out 100+ watts per pair for extended periods of time without ANY glow, and the idle dissipation is 6 watts! (600 V 10 mA). I ran it for 10 minutes alternating each minute between 2% distortion (100W) and clipping (125W). Nothing glowed, and nothing broke.
I am using the Edcor CXPP100-8-3.3K. I have seen these pass 200+ watts when used above 70 Hz (guitar amp). I plan to use this amp for several things. One is a full range HiFi amp where the B+ will be 600 volts and power at clip is 125W. Another use is a stereo guitar amp where more power is always good. Bigger tubes and more B+ can get there. Another use is a DJ style PA where the tube amp handles everything above 150 Hz or so. The Subwoofer duties are handled by TI 300 Watt class D chip amps.
In that board I squeezed 250 WPC from each pair of 35LR6's on 650 volts with a 2500 ohm load. At this power level there was a pale plate glow in a dark room. Not bad since I figure that each tube is eating about 60 watts!
All my amp building is currently on hold since I packed up Tubelab and moved 1200 miles. I won't have a lab for 6 months to a year. Before packing up everything I completed a single channel prototype of my "Killer". It started out as one of many LT Spice simulations, and wound up a working board. It uses a pentode LTP input/driver stage and sweep tube output stage. Both G1 and G2 are driven, solving the blown screen grid problem. Output tube plate to driver plate feedback is used to reduce output impedance.
The simulation and prototype used 6EJ7's and 13GB5's because they are plentiful, cheap, and the models are very good. I may choose different tubes for the final build. Does it work? Well the lowly $1 13GB5 can crank out 100+ watts per pair for extended periods of time without ANY glow, and the idle dissipation is 6 watts! (600 V 10 mA). I ran it for 10 minutes alternating each minute between 2% distortion (100W) and clipping (125W). Nothing glowed, and nothing broke.
I am using the Edcor CXPP100-8-3.3K. I have seen these pass 200+ watts when used above 70 Hz (guitar amp). I plan to use this amp for several things. One is a full range HiFi amp where the B+ will be 600 volts and power at clip is 125W. Another use is a stereo guitar amp where more power is always good. Bigger tubes and more B+ can get there. Another use is a DJ style PA where the tube amp handles everything above 150 Hz or so. The Subwoofer duties are handled by TI 300 Watt class D chip amps.
As far as I know, RCA 35LR6s are relabeled Sylvania parts, based on the plate structure and labeling scheme.
Since I only want a mere 100W/channel, I may back off and use some 6DQ5s instead of the 35LR6. The Antek guys swear that their XFMRs will deliver more power at 60Hz than 50 Hz, and this may be enough to supply the 6DQ5 filaments via the filament windings available on the Antek toroid. If I use a mosfet input stage, this would greatly simplify the power supply of this design.
P.S. I also thought of 6EJ7s for the input stage - great minds think alike...
Since I only want a mere 100W/channel, I may back off and use some 6DQ5s instead of the 35LR6. The Antek guys swear that their XFMRs will deliver more power at 60Hz than 50 Hz, and this may be enough to supply the 6DQ5 filaments via the filament windings available on the Antek toroid. If I use a mosfet input stage, this would greatly simplify the power supply of this design.
P.S. I also thought of 6EJ7s for the input stage - great minds think alike...
I built one of my SPP amps using a 100 VA Antek toroid. The Kill A Watt says that I am sucking 154 VA through it. It is mounted inside the box and gets pretty toasty but still lives on after 5 years. It is still far cooler than a equally overloaded Hammond in my SSE, which is on its tenth year. After several hours of continuous operation you can't touch the Hammond.
I chose the 6EJ7 because I have a bunch of them, and the LT Spice model works good. There are about a dozen tubes that can be plugged into the same socket too.
I have about 40 NOS Sylvanis 35LR6's but QC on these tubes was poor, so I don't want to put them into a rack mount amp that could see rough service. The mica spacers on several tubes are loose and I have already seen one with a loose plate and another with a loose control grid.
I have several other choices, so I will sort that out at build time. Power will likely be two Antek toroids if they ever get them back in stock. We used dual Antek setup in the 125 WPC amps we made in the past. Each is wired to make a 325 volt supply and the two supplies are wired in series so that the driver and screen grids run from the 325 volt supply, and the plates get 650 volts. This gives you 4 X 6.3 volts for heater options.
I chose the 6EJ7 because I have a bunch of them, and the LT Spice model works good. There are about a dozen tubes that can be plugged into the same socket too.
I have about 40 NOS Sylvanis 35LR6's but QC on these tubes was poor, so I don't want to put them into a rack mount amp that could see rough service. The mica spacers on several tubes are loose and I have already seen one with a loose plate and another with a loose control grid.
I have several other choices, so I will sort that out at build time. Power will likely be two Antek toroids if they ever get them back in stock. We used dual Antek setup in the 125 WPC amps we made in the past. Each is wired to make a 325 volt supply and the two supplies are wired in series so that the driver and screen grids run from the 325 volt supply, and the plates get 650 volts. This gives you 4 X 6.3 volts for heater options.
If you have a bunch, the 6EJ7 can be attractive that way. What got my attention was the high plate voltage rating (compared to similar pentodes) and the high transconductance. I'm not totally engineer in my genetic sequence, so I also like things that look nice. The Euro-made 6EJ7s with the silver mesh shield look pretty swank. I've seen some (US-made) that just look kinda dirty with smoked innards, and one or two with a perforated metal outer shield.
They also seem to be electrically "different" that the US made tubes. I worked in a TV repair shop when TV's with the 6EJ7 and 6EH7 hit the market. At that time all of the US branded tubes were relabeled European tubes and the TV's were on the verge of instability. In fact some Philco TV's would exhibit IF oscillation while trying to receive a weak channel 5 station in the presence of a very strong adjacent channel 6. Tube rolling was the only solution. The US made tubes seemed to eliminate the instability, but some TV's did not like them.
I have about 100 used 6EJ7's and maybe 10 NOS tubes. Some are microphonic. I don't think any are bad enough to preclude their use as a driver in a power amp. I don't think I would use them in a phono stage though.
This was the late 60's. Zenith was still extoling the virtues of the hand wired TV chassis and other manufacturers were finding out that Phenolic PC boards don't live long in a hot TV chassis in a dusty humid house.
We were a Philco (owned by Ford motors at the time) authorized repair shop. 75% of our work was Philco TV's. The layout of the IF amp PCB was poor even by my teenage design standards. It was a single sided G10 board with routed power and ground. I don't think there were any stopper resistors, probably one or two bypass capacitors, and they were tubular Mylar. The 6EJ7 had the highest Gm seen at the time....of course it will oscillate.
The GE TV's of the time had their unique Lexan high voltage cage. Add some generous dust, some extreme south Florida humidity, maybe a few ants or roaches, and you have the recipe for a meltdown. Once the HV box got dirty enough to arc, the plastic would carbon track, creating a carbon short from the output of the flyback to ground, which was fed by the B+ supply. Either the flyback caught fire, or the Lexan box melted! They went back to metal a year or two later.
We were a Philco (owned by Ford motors at the time) authorized repair shop. 75% of our work was Philco TV's. The layout of the IF amp PCB was poor even by my teenage design standards. It was a single sided G10 board with routed power and ground. I don't think there were any stopper resistors, probably one or two bypass capacitors, and they were tubular Mylar. The 6EJ7 had the highest Gm seen at the time....of course it will oscillate.
The GE TV's of the time had their unique Lexan high voltage cage. Add some generous dust, some extreme south Florida humidity, maybe a few ants or roaches, and you have the recipe for a meltdown. Once the HV box got dirty enough to arc, the plastic would carbon track, creating a carbon short from the output of the flyback to ground, which was fed by the B+ supply. Either the flyback caught fire, or the Lexan box melted! They went back to metal a year or two later.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.