@frank754:
Here is a design I found that, even though it's a modulator, and uses different VTs, that it could be worked up into what you've been wanting: a Class AB2, 807 design.
This is basically a Williamson, and should be quite easy to modify into a good sounding amp. He's got most of it right as it stands right now.
Here is a design I found that, even though it's a modulator, and uses different VTs, that it could be worked up into what you've been wanting: a Class AB2, 807 design.
This is basically a Williamson, and should be quite easy to modify into a good sounding amp. He's got most of it right as it stands right now.
Thanks, Miles, for the info on this, much appreciated.
I have a few questions on your Vixen design, if you have a minute (or when you do, no hurry), ...
What's the purpose of the R21 & R22 pots going into the grids of the 6SN7, to fine tune the voltage on each grid so that they can be set at exactly -29.5v each time a new tube is installed?
Can the 807's be random tubes, not matched, or do R21 & R22 balance both the 6SN7 and 807's simultaneously?
L1 & L2: 10 turns of 18 ga enameled magnet wire on a 7/16 ID hollow form, so the OD would be 1/2"?, etc. = just about 1 microhenry (my guess according to the info in my 1957 Allied Electronics Data Handbook). Do these help attenuate some type of annoyance?
On the power supply, what are Z1 & Z2 on the mains side, and do the integrated bridges provide any benefit over 4 diodes?
I found the Fairchild 1084 (but it comes in all these versions:
FAN1084D15X FAN1084DX FAN1084M33X FAN1084MC33X FAN1084MCX FAN1084T FAN1084T15 - any one in particular or does it matter?). How does a DC regulated filament supply help, aside the AC hum from the filament wires to the tube sockets?
And the real fine-tuned 3-tube screen bias supply, can't this be done alternatively with zener diodes, and if the bias were to err a bit, would it be best to have it slightly higher than 270 to protect the tubes, pretty much what do you think the tolerance could be, 10v higher, 5v less?
Just trying to get an expert's feel here, I haven't worked to closely with these types of tolerances.... thanks again.
I have a few questions on your Vixen design, if you have a minute (or when you do, no hurry), ...
What's the purpose of the R21 & R22 pots going into the grids of the 6SN7, to fine tune the voltage on each grid so that they can be set at exactly -29.5v each time a new tube is installed?
Can the 807's be random tubes, not matched, or do R21 & R22 balance both the 6SN7 and 807's simultaneously?
L1 & L2: 10 turns of 18 ga enameled magnet wire on a 7/16 ID hollow form, so the OD would be 1/2"?, etc. = just about 1 microhenry (my guess according to the info in my 1957 Allied Electronics Data Handbook). Do these help attenuate some type of annoyance?
On the power supply, what are Z1 & Z2 on the mains side, and do the integrated bridges provide any benefit over 4 diodes?
I found the Fairchild 1084 (but it comes in all these versions:
FAN1084D15X FAN1084DX FAN1084M33X FAN1084MC33X FAN1084MCX FAN1084T FAN1084T15 - any one in particular or does it matter?). How does a DC regulated filament supply help, aside the AC hum from the filament wires to the tube sockets?
And the real fine-tuned 3-tube screen bias supply, can't this be done alternatively with zener diodes, and if the bias were to err a bit, would it be best to have it slightly higher than 270 to protect the tubes, pretty much what do you think the tolerance could be, 10v higher, 5v less?
Just trying to get an expert's feel here, I haven't worked to closely with these types of tolerances.... thanks again.
frank754 said:
R21 and R22 allow setting the current for each 807 to be identical. That would be 0.48V across the 10R current sampling resistors at the 807 cathodes. (48mA Plate + Screen current) Has nothing to do with "balancing" the 6SN7s, which won't be bothered by any reasonable imbalance since they're cathode followers, and have a wide latitude for Q-Point. They can't be varied all that much anyway. Balancing the 807 currents helps reduce any xfmr DC magnetization, and doesn't require specially matched 807s.
Plate stoppers. The coils break up any resonant circuits formed from device capacitance, xfmr stray capacitance, lead and xfmr leakage inductance. The 100R resistors de-Q the coils so that they don't resonate themselves. The 1.0uH inducatnce of the coils is of no consequence at audio, but helps to prevent RF parasitic oscillation.
Your standard, 120Vac MOVs. These help protect against any HV impulses that try to ride in on the AC mains. Helps to prevent poofage of the SS diodes. When installing these, they must come after the fuse, not in front of it. MOVs sometimes have a nasty habit of staying on once switched on by a HV transient. If that current isn't interrupted right away:
Other than to take up less circuit board real estate, not that I'm aware of.
I used a FAN1084T. It would seem that the different parts numbers refer to different packaging schemes. Which one you choose, make certain that the three leads are: IN, OUT, and ADJ. These regulators may have their output voltages set. The ones with: IN, OUT, GND are fixed voltage regulators. These are used mainly for computer PSs (these would be the ones ending in 15, or 33). DC heater power helps with AC hum, and also allows operation without having to elevate the heaters above ground to turn off the "diode" formed from the hot aluninum oxide insulation (goes sort of semiconductor at red heat) and the nickle cathode sleeve.
Yes it can, however Zeners are very noisey since they conduct a reverse current. Otherwise, you could certainly solid state that regulator. I didn't do that mainly because I already had the VTs on hand, but no high voltage transistors. All-in-all, the VT regulator works just great. It'll hold the output to within 0.8Vdc for 40Vdc rail variations, and has not drifted at all since I set them up over a year ago. The voltage will be a bit higher since the 807s required 1.5K screen stoppers to completely eliminate RF parasitics (60KHz damped oscillation on signal peaks) so if the voltage at the 807 screens is 270Vdc, the regulator will be slightly higher. Although the normal Vpk at the 6AQ5 series pass tube is only 95Vdc anyway, so it's not in any danger on that regard.
Besides, it scores glowey bottle kewelness points.
Hey Miles,
I too have been gathering schematics for the 807 (and 6L6 and 5881) in the hopes of putting my stash to use someday. Any chance your design would be a good match for my inefficient ESL 57s?
Up until now I'd been favoring the Grommes 216BA circuit (uses 6BG6 - octal 807, I believe):
http://www.triodeel.com/grm216ba.gif
Thanks for your time and expertise on this.
I too have been gathering schematics for the 807 (and 6L6 and 5881) in the hopes of putting my stash to use someday. Any chance your design would be a good match for my inefficient ESL 57s?
Up until now I'd been favoring the Grommes 216BA circuit (uses 6BG6 - octal 807, I believe):
http://www.triodeel.com/grm216ba.gif
Thanks for your time and expertise on this.
this is what the 1625 was designed for -- B17, B29 etc.
An externally hosted image should be here but it was not working when we last tested it.
CarlyBoy said:
Now, that, I don't know. I've used it with electromagnetic speeks only, not electrostatic, so I can't say.
That's an assumption that I'd be careful about. That Grommes circuit operates the plate and screen at the same voltage: 440Vdc. The 807 rates its screen at a max of 300Vdc. The 6BG6 may be quite similar to the 807, but it was designed for TV horizontal deflection PA use, and may have been "hardened" to withstand higher voltages. I don't ever recall 807s or 6L6s being used for horizontal deflection duty. It may poof if the screen is overvolted by some 140Vdc. (Not saying it will, but there was a limit of 300Vdc set on the screen for a reason.)
If that's the one you're going with, I'd use 5963s instead of 12AU7s. They're more linear, but have otherwise similar specs.
I attempted to operate a boatanchor like that on 40 meters in the early 60's. I was trying to use a Hallicrafters SW set for RX. Didn't make many contacts.
I got a large number of surplus tubes a few years ago including several thousand broken tubes. I dissected several 1625, 807, 6BG6, 6L6G and 6L6GB tubes. I believe that within a given manufacturer and date code run these tubes were all the same. There is some different construction between the different manufacturers tubes, especially the 807's and 1625's. There were also a lot more brands of 807's made to support the war effort.
The 6BG6's that I tested (Sylvania) had no problem handling 420 volts on the screen, but the plate got hot and bothered at about 20 watts.
6L6 tubes would never live in a TV set because of the high peak voltage on the plate during horizontal retrace. It would arc from pin 3 to pin 2 at the socket. This is often the limiting factor in getting high power from the common audio tubes, especially in guitar amp service where the load may not be matched and the amp is driven to clipping. 807's would work in sweep tube service, but by the time that they were commonly available (after WWII) better tubes had been developed (6BQ6, 6DQ6, 6CD6 etc)
As tube production wound down many manufacturers stuffed whatever guts they could make into some tube envelopes to satisfy government contracts. There were rumors that Sylvania was even rewiring the bases of 6AV5's (an indirectly heated sweep tube pentode) and selling them as 6B4G's (A DHT triode). This is actually true! I have seen and tested the tubes, and they do work! Angela Instruments has some for sale. Along the same lines Sylvania stuffed some 7027A guts into some 6BG6 glass, and sold them as 6BG6's. I have tested a pair, and they can handle some extreme abuse. The guts look a lot like the Sylvania 6L6STR that is highly sought by guitar amp users for its robustness. If you need 6BG6's that can crank to 11, use these:
http://www.vacuumtubes.com/6BG6.html
I used to have one of those HF boatanchor surplus transmitters too, that my dad, a WWII vet brought back. Unfortunately I sold 90% of my stash, 1000+ tubes and about 20 cigar boxes of components, probably 50 pots and some test equipment (some Heathkit stuff I built as well back in the 70's, and a working DuMont scope too, for $200 about 10 years ago, was a bad move.
Miles, thanks for your help, I'm 90% sure of making your amp my next project in maybe 4-6 weeks, after I get my current stuff working. I'm on the road this week, and will have a few more very brief questions maybe this weekend.
This is a great thread, and some excellent info and opinions from a wide range of people so far.
Miles, thanks for your help, I'm 90% sure of making your amp my next project in maybe 4-6 weeks, after I get my current stuff working. I'm on the road this week, and will have a few more very brief questions maybe this weekend.
This is a great thread, and some excellent info and opinions from a wide range of people so far.
Further to higher than recommended screen voltages, how about this one:
http://www.ampslab.com/SCHEMATICS/Grommes215.gif
Did the folks at Grommes know something? Did they source some sort of super 807? Would be interesting to know how many units were sold and their service history...
http://www.ampslab.com/SCHEMATICS/Grommes215.gif
Did the folks at Grommes know something? Did they source some sort of super 807? Would be interesting to know how many units were sold and their service history...
I've had great experience with these; using them in Eico HF87 and Fairchild 260 rebuilds. And at that price... wow.
OK, this is for Miles Prower, if you are around, I'm starting to source & scope out the parts for your Vixen plan, and have a few questions.
For the 6J5, the metal tube version should be OK, since it's the first stage?
I've found the MPSA42 transistors at Allied, just 14 cents, but the 1N941A Zener diodes seem a bit rare, is there a common substitute? That exact number doesn't come up on Allied, I try to keep all my suppliers down to 3, usually Antique Electronics, Allied, and Angela.
I've already done a search and sent an email on the 1 microhenry choke for the plates of the 807's, if this doesn't pan out, can I just wind them on 1/2" dowels using 22ga wire and put the resistor outside?
As for the expensive Auricaps, can I use instead use 400 volt metallized polypropylene caps (ASC brand) without too much ill effect?
I'm currently planning to build a monoblock, but with your full regulator schematic for the Vsgsg. I have probably 30 6CB6A's (wondered when they would ever come in handy, and probably 5 or 6 6AQ5's, if these are not "A"'s should they still be ok?)
For the power supply, since I'm planning the use 1625's and not 807's, I do have a separate trafo for the 12.6 v tubes, which I won't rectify (if that's not important), since it's the last stage, perhaps I can also use 12SN7's and put them on the same filament transformer as well, for the stage just prior.
I can run the bias power supply tubes from the 6.3v supply of the main 300-0-300 (for the 6CB6A and 6AQ5), since it looks like you already have a "riser" for hum control from your 110k and .22 cap in this place.
That would leave just the sensitive 6J5 and 6SL7 to get the DC filament supply, if you deem that's enough.
Aside from using the Fairchild "FAN" part, and an 8 volt filament supply, is there an easier way I can get by with this. That's my major sticking point. Maybe just run a DC bridge with your caps and skip the "FAN" thing, or even with AC on the preamp stage filaments, would that be a crime if properly elevated? I can see the perfectionist viewpoint, but a lot of the old amp circuits didn't even have half the features that I'm already including, like your grid regulator (I've already scoped out the multiturn Cermet pot) and transistor/zener cathode supply for the 6SL7.
For the 6J5, the metal tube version should be OK, since it's the first stage?
I've found the MPSA42 transistors at Allied, just 14 cents, but the 1N941A Zener diodes seem a bit rare, is there a common substitute? That exact number doesn't come up on Allied, I try to keep all my suppliers down to 3, usually Antique Electronics, Allied, and Angela.
I've already done a search and sent an email on the 1 microhenry choke for the plates of the 807's, if this doesn't pan out, can I just wind them on 1/2" dowels using 22ga wire and put the resistor outside?
As for the expensive Auricaps, can I use instead use 400 volt metallized polypropylene caps (ASC brand) without too much ill effect?
I'm currently planning to build a monoblock, but with your full regulator schematic for the Vsgsg. I have probably 30 6CB6A's (wondered when they would ever come in handy, and probably 5 or 6 6AQ5's, if these are not "A"'s should they still be ok?)
For the power supply, since I'm planning the use 1625's and not 807's, I do have a separate trafo for the 12.6 v tubes, which I won't rectify (if that's not important), since it's the last stage, perhaps I can also use 12SN7's and put them on the same filament transformer as well, for the stage just prior.
I can run the bias power supply tubes from the 6.3v supply of the main 300-0-300 (for the 6CB6A and 6AQ5), since it looks like you already have a "riser" for hum control from your 110k and .22 cap in this place.
That would leave just the sensitive 6J5 and 6SL7 to get the DC filament supply, if you deem that's enough.
Aside from using the Fairchild "FAN" part, and an 8 volt filament supply, is there an easier way I can get by with this. That's my major sticking point. Maybe just run a DC bridge with your caps and skip the "FAN" thing, or even with AC on the preamp stage filaments, would that be a crime if properly elevated? I can see the perfectionist viewpoint, but a lot of the old amp circuits didn't even have half the features that I'm already including, like your grid regulator (I've already scoped out the multiturn Cermet pot) and transistor/zener cathode supply for the 6SL7.
frank754 said:
Certainly. That's what I'm using: Westinghouse JAN metal 6J5s. Insisting on the glass version is just a bit of audiophoolery.
1N941As aren't Zeners. These are common, small signal silicon diodes. I got a 50 pack of them from Radio Shack for about a buck. I can't imagine why you're having such a hard time finding these, but any common, silicon, small signal diode would work just fine here. Even Zeners connected in the forward direction would be OK, just so long as they have the 0.7Vdc forward voltage.
Don't waste any money buying these chokes. It is extremely easy to wind them yourself around a 7/16ths inch drill bit. You could use #22 wire, but #18 will give better mechanical support to the coil/damping resistor combination.
Most likely, since this is a gNFB amp. I like AuriCaps myself, but the ones you describe ought to be close. However, don't use a 400V capacitor for the local NFB DC blocking capacitor (from the 807 plates to the 6SN7 grids) that one needs to be at least a 600V capacitor, otherwise, it could poof and take out the driver and final.
The only difference between a 6AQ5 and a 6AQ5A is that the latter has what's called a "controlled heater warmup characteristic". This is important only in those situations where all filaments are connected in series. Otherwise, it makes no difference.
Yes, but just make certain that you have some provision for hum balancing the AC heater voltage.
I would advise against this. Since the amp uses a SS power supply, the HV appears within a couple of seconds. In this design, all heaters are to be up to operating temp before the HV is switched on. I don't know what'll happen if you delay the heater voltage to the regulator. It wasn't designed to be operated that way. The "riser", as you call it, has nothing to do with hum control. It's required so that the Vhk ratings of the VTs aren't exceeded so that the cathode voltage doesn't arc over to the heater.
AC on the filaments probably won't present too much of a problem if you have hum balancing. DC isn't quite so critical here as it would be on a much lower level signal. I included DC heater power as a further optimization. It'll work just fine without it, even if there is a bit more hum involved. They got away without it in "the day" when high current SS diodes and IC regulators weren't available. That's preferable to unregulated DC since part of what the IC regulator does is suppress residual ripple. AC is better than "dirty" DC for filaments since it doesn't have the harmonic content of unregulated DC, and that harmonic content is more sonically detrimental than AC.
The 1N941A is a rather obscure and extinct zener diode. This is however not the number that you are looking for. You want a 1N914A which is about the most common diode on the planet.
If you can't find those, use the second most common diode on the planet, the 1N4148.
OK, thanks, I think I've gotten the list together now. Also appreciated tubelab mentioning using 1N914A's or 1N4148's, must have been a typo on the parts list (in the 6SL7 cathode tail)
As far as the topology of the power supply, I'm planning to get a separate 12.6VCT 3 amp transformer for that, and run all the tubes from that on AC (with hum control divider resistors), both the 6.3 & 12.6v tubes, and being a monoblock the total filament current just exceeds 2 amps according to the tube manual specs. Four 12v tubes (two 1625's, a 12SN7GT and a 12SL7GT), and I will balance the remaining 3 tubes (6J5, 6AQ5A, 6CB6A), from one side or other of the center tap. I already have a nice Sylvania JAN 12SL7GT with a US Navy anchor symbol painted on the base that tests good.
This way I'll have the protection of a separate standby switch for the HV, and use the 6.3 winding from the main xfrmr just for pilot lights, etc...
I do plan to use a 5U4GB instead of diodes for the rectifier anyway, since I have at least two extra of these kicking around that test good, plus it'll look good with a extra big tube. Maybe in that case I can skip the standby switch? I already have a 7H 150ma choke, is that good enough, or should I splurge the extra $29 for the 8H, as it comes enclosed to match the power transformer?
Being a monoblock it seems I may not need 250ma from the 300-0-300 trafo (from what I gather you specified that for a 2-channel unit), I'm thinking now that the Hammond 272DX (125ma) should handle it, a smaller unit. (or 272FX, with 150ma)
A few final quick questions: the cermet pot, is one of those tiny blue 1/4-1/2watt jobs good enough, or should I get a multiturn precision pot (2-3 watts or more). I'm wondering why those tiny cermets say just 1000 hours life in the catalog specs?
And the feedback coax, you mention RG-75, is any 75 ohm video coax ok, like RG-6 or RG-59U?
Other than that it looks good to go... appreciate all the advice 🙂
Getting ready to order 🙂
As far as the topology of the power supply, I'm planning to get a separate 12.6VCT 3 amp transformer for that, and run all the tubes from that on AC (with hum control divider resistors), both the 6.3 & 12.6v tubes, and being a monoblock the total filament current just exceeds 2 amps according to the tube manual specs. Four 12v tubes (two 1625's, a 12SN7GT and a 12SL7GT), and I will balance the remaining 3 tubes (6J5, 6AQ5A, 6CB6A), from one side or other of the center tap. I already have a nice Sylvania JAN 12SL7GT with a US Navy anchor symbol painted on the base that tests good.
This way I'll have the protection of a separate standby switch for the HV, and use the 6.3 winding from the main xfrmr just for pilot lights, etc...
I do plan to use a 5U4GB instead of diodes for the rectifier anyway, since I have at least two extra of these kicking around that test good, plus it'll look good with a extra big tube. Maybe in that case I can skip the standby switch? I already have a 7H 150ma choke, is that good enough, or should I splurge the extra $29 for the 8H, as it comes enclosed to match the power transformer?
Being a monoblock it seems I may not need 250ma from the 300-0-300 trafo (from what I gather you specified that for a 2-channel unit), I'm thinking now that the Hammond 272DX (125ma) should handle it, a smaller unit. (or 272FX, with 150ma)
A few final quick questions: the cermet pot, is one of those tiny blue 1/4-1/2watt jobs good enough, or should I get a multiturn precision pot (2-3 watts or more). I'm wondering why those tiny cermets say just 1000 hours life in the catalog specs?
And the feedback coax, you mention RG-75, is any 75 ohm video coax ok, like RG-6 or RG-59U?
Other than that it looks good to go... appreciate all the advice 🙂
Getting ready to order 🙂
frank754 said:
If you go with a VT rectifier, you won't need a separate HV switch, and a 7.0H choke will work. 7.0H is close enough, and the main reason I went with the 8.0H choke is because that's what was in the junk box.
Yes, the bigger xfmr is intended for a two-channel unit. The 125mA xfmr will be OK for a monoblock. The output voltage may be a bit lower since the 5U4GB has a higher forward voltage than does a silicon diode. It shouldn't affect things adversely since VT circuits are quite a bit more forgiving than are SS.
I'm using those small, blue multi-turns for the screen regulator. They work just fine, and I wouldn't worry too much about that 1000hr rating since these are set it and forget it unless it becomes necessary to replace a 6CB6 or 6AQ5, and then it's just a question of touching it up a bit. The ones I have haven't drifted even a little bit since they were set up over a year ago. The voltage regulator is quite stable. It's the constant twiddling of the resistor that wears 'em out.
Now, here's a problem. When using the RG-75, and with the way I constructed it, I found that a feedback compensating capacitor wasn't necessary. Obviously, the cable capacitance was all that was required. If using a different construction, and different cable, that may no longer be the case. You may have to o'scope and experiment to find out what that compensating capacitor needs to be. That's one of those "unknowns" that can be determined only by observation.
Also, I would get a separate heater xfmr for the regulator, since this requires 150Vdc for Vhk protection. Putting all the heaters at 150Vdc will exceed Vhk ratings on the other stages. If that power xfmr has a 6.3Vac winding, then that's what I'd use it for since you're gonna go with VT rectifiers and won't be needing a HV standby.
Great, thanks again. That very last item (the 150v protection) is something I wouldn't have noticed, as far as being incompatible with the rest of the circuit. The main xfmr has a 6.3 which I can solely dedicate to the VR tubes, so I guess that irons it out.
Will post back in a week or so with progress, once I get the parts.
Am trying to go for a good look as well, with an 8x15" brass anodized aluminum frame (1/2 of a shiny gold-looking 8x30 door kick plate from Lowes at $18 for the whole sheet), and mount it on 1/2" x 2 1/2" oak trim boards (rectangular frame). Guess I'll put all the switches & jacks on top to make it easier.
Again, I appreciate your time, this seems to be the makings of my most enjoyable project yet...
Will post back in a week or so with progress, once I get the parts.
Am trying to go for a good look as well, with an 8x15" brass anodized aluminum frame (1/2 of a shiny gold-looking 8x30 door kick plate from Lowes at $18 for the whole sheet), and mount it on 1/2" x 2 1/2" oak trim boards (rectangular frame). Guess I'll put all the switches & jacks on top to make it easier.
Again, I appreciate your time, this seems to be the makings of my most enjoyable project yet...
Miles Prower said:
The article implies that the amplifier is capable of 100% modulation of a 200 watt AM RF amplifier, and for that you need only 100 watts of audio.
More precisely, an RF amp that sucks 200 watts DC, for an output of maybe 180 watts or so.Circlotron said: