4x 6550 bass amp design

[andrew_k]

I really like how you used 2 x 230VAC windings then used the screen supply to elevate the plate supply.

Lifted from HIWATT DR201

would it be better to use the choke right after the BR as opposed to between the PI and the screen node so that you still get choke regulation on the screen tap? Or will the kick back from the choke damage the BR if it doesn't have a filter cap or resistor upstream of it?

If you have a big enough choke to supply the current requirements, sure. I only had a small one which is already powering 8 triodes and may be called upon to power more in the future.

what's the purpose of the 820R/330K divider directly after the BR?

820Ω = a little sag
330K = drain the caps when power is off. 670V hurts.

And why two seperate heater windings for power and preamp tubes? Any advantages to this?

Necessity. The PT has two 4A 6.3V windings, the two power valves alone consume 3.3A.

[Steve Conner]

@Steve Connor, I noticed you recommended to use a doubler in your post above. Won't going that route only get you 1/2 the current? Or would that not matter as it would still be 100 watts, just with a different voltage/current ratio?

Yes. His OT has too high an impedance for a quad of 6550s running off 400V, so I propose making the power with twice the voltage and half the current.

Math time: We have a 5k OT, so that's about 1.25k from the CT to an end. So the maximum peak power you could ever have is about 800 squared divided by 1250, which is 512 watts, and the mean output on a sine wave (the annoyingly named "RMS power") is half of that, 256 watts.

In practice probably about 180, since I neglected the tube drop and B+ sag.

With the 400V B+ you're only getting one quarter of this or less: something like 60 watts, and you probably get the same output with two tubes as with four. If you mismatched the load one step down, it might wheeze its way over 100W.

The theoretical current draw from B+ on the maximum unclipped sine wave output is (800/1250)/1.6 = 400mA. If you were using a doubler that would look like 800mA to the PT.

But isn't a Twin PT meant for a tube rectifier? So you can use it with a bridge, and the better utilization factor gives you more or less extra free current. (the bridge uses both halves of the winding all the time, the FWCT only uses one at a time)

[Wilder Amplification]

Yes. His OT has too high an impedance for a quad of 6550s running off 400V, so I propose making the power with twice the voltage and half the current.

Math time: We have a 5k OT, so that's about 1.25k from the CT to an end. So the maximum peak power you could ever have is about 800 squared divided by 1250, which is 512 watts, and the mean output on a sine wave (the annoyingly named "RMS power") is half of that, 256 watts.

In practice probably about 180, since I neglected the tube drop and B+ sag.

With the 400V B+ you're only getting one quarter of this or less: something like 60 watts, and you probably get the same output with two tubes as with four. If you mismatched the load one step down, it might wheeze its way over 100W.

The theoretical current draw from B+ on the maximum unclipped sine wave output is (800/1250)/1.6 = 400mA. If you were using a doubler that would look like 800mA to the PT.

But isn't a Twin PT meant for a tube rectifier? So you can use it with a bridge, and the better utilization factor gives you more or less extra free current. (the bridge uses both halves of the winding all the time, the FWCT only uses one at a time)

Thanks a bunch Steve. Just so you know my background...I've been an electronics guru for about 18 years, with 13 of that being tube amp experience. I'm sure you can tell that I'm not a "kit assembler" and actually understand the theory and know how to read a schematic as well as know full and well how an amplifier works down to the component level. However, most of my experience has been with clones and am now branching out into the engineering and design sector of things, which is the 'math behind the mystery' that I never learned.

I understood all of the math you gave except for one thing - the (800/1250)/1.6 = 400mA. I understood the first part and how you arrived at the answer, however I couldn't figure out where the 1.6 comes into play. Is that a push pull amp math constant of some sort? Or is that a variable derived via some other means?

Also...rethinking my power supply design...considering going up 100 on the plates and 50 on the screens for a 700/350 supply on 2 x 6550/KT88s for a cleaner 100 watts. My design will be running a 5K p-p load.

[andrew_k]

Also...rethinking my power supply design...considering going up 100 on the plates and 50 on the screens for a 700/350 supply on 2 x 6550/KT88s for a cleaner 100 watts. My design will be running a 5K p-p load.

700V is above max plate voltage for 6550

[Steve Conner]

Hi Wilder,

1.6 is the peak-to-average ratio of a rectified sine wave.

800/1250 gives us the maximum theoretical peak current that either push-pull side could draw.

So we know that at full sine wave power, the current waveform at the OT center tap is a rectified sine wave whose peak value must be less than 800/1250.

Therefore, the average current drawn from the power supply must be (800/1250)/1.6 or less.

In practice you can overdrive the thing, mashing the output into a square wave, which when rectified has a peak-to-average ratio of 1. So a really bombproof amp, designed to handle some abuse, would use 640mA as the rating.

Maximum plate voltage ratings for tubes are pretty negotiable, lots of designs exceed them and get away with it. It's the screen voltage that can bite you.

Just so you know my background, I design things full of microprocessors for a living and spend most of my time programming. Chips for business, tubes for pleasure.

[Wilder Amplification]

700V is above max plate voltage for 6550

Tell Ampeg that.

In all honesty I'll probably stay at 600 unless I plan to make an amp exclusively for KT88s.

Just so you know my background, I design things full of microprocessors for a living and spend most of my time programming. Chips for business, tubes for pleasure.

LOL...before I asked a question that I probably should've known the answer to I figured I'd just explain my background so that you know I'm not a COMPLETE hack. But thanks for that...I think I'll just go and stick my nose back in the books before I attempt anymore builds. *sarcasm*

Never did get near as far as where you're at electronically. Once I discovered tubes I was quite happy and content where I'm at, although I could use some knowledge of digital electronics at some point.

I hear ya on the screens...most people write it off to the CP tubes not being able to take the plate voltage when in reality it's the screens that are the problem (hence why Marshalls are notorious for shorting screens/taking out screen resistors) and as such your plate voltage is limited to just slightly above that on a single rail supply, which is why I opted to get into dual rail supplies...more so for longer tube life. And one thing that comes with dual rail supplies and bigger OT loads...more power per pair of tubes! So why not take advantage of that?

[Steve Conner]

Once I discovered tubes I was quite happy and content where I'm at, although I could use some knowledge of digital electronics at some point.

Noooo! That's the last thing the world needs, more digital amps... Logic gates in the footswitch circuit are bad enough

Tubes have a wonderful "less is more" thing going on. To an EE type, it's more like "bad is the new good". The old tube amp circuits measure horribly in the lab and then infuriate EEs by sounding great anyway.

[Wilder Amplification]

Noooo! That's the last thing the world needs, more digital amps... Logic gates in the footswitch circuit are bad enough

Tubes have a wonderful "less is more" thing going on. To an EE type, it's more like "bad is the new good". The old tube amp circuits measure horribly in the lab and then infuriate EEs by sounding great anyway.

Never will I resort to logic gates for switching. I use a high sensitivity relay for that...the Omron G5V-2 series that only draw 30mA. The heater supply runs them nicely.

Now going by your equations, for my design with a 600V B+ and a 5K p-p load (2 x 6550s @ 100 watts) -

600 squared = 360,000

360,000 / 1250 = 288 watts peak

(288 watts / 600 Volts)/1.6 = 180mA peak current

I spec'ed the plate winding at 400mA which I now see is more than enough for an amp that will see most of its duty running clean. For a "bombproof" design in the case of my amp the calculation would be about 480mA, which I'd round that off to about 500mA. So I may just re-spec it at 500mA so the tranny will be happy in the event it gets pushed into max clipping.

Makes me wonder how these old Marshalls did it. Lots of 100 watters had 500V B+ with a 1.7K p-p load, which given 425 plate - center tap -

500 squared = 250,000

250,000 / 425 = 588 peak watts

(500 / 425) / 1.6 = 735mA clean
500 / 425 = 1.18 A full clip

Doesn't quite equate...seeing as how my Heyboers in my 100 watt hot rod 2203 clone are only rated to source 400mA off the HT winding.

[Steve Conner]

Well, as the B+ voltage goes down, the tube drop gets even more important. Tubes aren't perfect switches, even when turned on as hard as possible they won't conduct fully until they have 50 or 100 or even 150V across them. 100 isn't large compared to 800, but it's a good proportion of 400. And of course the lower B+ needs more current to make the same wattage, and that means a higher tube drop still.

Having said that your figures seem reasonable. When I've measured 50 watt tube amps, they draw about 300mA current from B+ for maximum clean power, and more for full clip. So double that for a 100W amp, and that's 600, not far off your 735.

In short, the equations I gave are certified for ballpark use only. And the transformer in your 100W amp is too small, though it's common in all audio amps, tube and transistor, to underrate the PT down to only 70% of the theoretical VA rating.

[Wilder Amplification]

Well, as the B+ voltage goes down, the tube drop gets even more important. Tubes aren't perfect switches, even when turned on as hard as possible they won't conduct fully until they have 50 or 100 or even 150V across them. 100 isn't large compared to 800, but it's a good proportion of 400. And of course the lower B+ needs more current to make the same wattage, and that means a higher tube drop still.

Which explains why they offer a much better "swing" when operated off a dual rail supply at a much higher Va with a higher plate load than they would be on a single rail supply and a lower plate load.

Having said that your figures seem reasonable. When I've measured 50 watt tube amps, they draw about 300mA current from B+ for maximum clean power, and more for full clip. So double that for a 100W amp, and that's 600, not far off your 735.

In short, the equations I gave are certified for ballpark use only. And the transformer in your 100W amp is too small, though it's common in all audio amps, tube and transistor, to underrate the PT down to only 70% of the theoretical VA rating.

Actually my B+ on that amp is 460, which brings things even closer to your 600mA calculation...actually came out to 636mA at full clean signal but 1.08A at full clip. However...this PT is supposed to be a clone of the Dagnall 1203-80 so it was probably purposely underrated to get it to sag under load in order to justify running a load as low as 1.7K for 4xEL34s without sending them into overcurrent at full clip. At least that's my "theory" anyway.

Now as far as for my 600V B+ dual rail monster, 450V @ 500mA should suffice?

[Steve Conner]

Now as far as for my 600V B+ dual rail monster, 450V @ 500mA should suffice?

Something like that... It seems more complicated than it needs to be. You can just take the 300V off the midpoint of your two series connected filter caps, no need for the separate winding you've drawn.

This is what I originally meant, you can take a PT that gave 400-something with a tube rectifier, and connect it to a bridge rectifier, giving you output from the centre tap at the original voltage, plus a new output at twice the voltage. That would normally be considered too high at 800-900V, but as you've been saying, I don't see why it wouldn't work. Just don't forget the rubber soled shoes when playing

[Wilder Amplification]

Something like that... It seems more complicated than it needs to be. You can just take the 300V off the midpoint of your two series connected filter caps, no need for the separate winding you've drawn.

This is what I originally meant, you can take a PT that gave 400-something with a tube rectifier, and connect it to a bridge rectifier, giving you output from the centre tap at the original voltage, plus a new output at twice the voltage. That would normally be considered too high at 800-900V, but as you've been saying, I don't see why it wouldn't work. Just don't forget the rubber soled shoes when playing

I had asked about that in a few posts above. Was originally considering the FWB doubler and taking the screen voltage off the middle junction but was concerned with power supply dragging when the screens start pulling current. Would spec'ing the current source of the HT winding higher make this a non-issue? Or is that not something I need to concern myself with?

[Steve Conner]

The current draws just add up, and the plate current one will dominate, so you probably can ignore the screen current getting drawn from the 300V node. Stick another 50mA onto the rating, maybe...

You have a point, though. High plate current draw from the 600V node will drag down the 300V supply, more so than if you used two separate windings.

I have an old PT somewhere from a FT560 ham transmitter, that I'm saving for a project like this. The transmitter made 300W from four TV sweep tubes, and the PT has about a half dozen different HT windings and about 12A of heater current.

[Wilder Amplification]

The current draws just add up, and the plate current one will dominate, so you probably can ignore the screen current getting drawn from the 300V node. Stick another 50mA onto the rating, maybe...

You have a point, though. High plate current draw from the 600V node will drag down the 300V supply, more so than if you used two separate windings.

I have an old PT somewhere from a FT560 ham transmitter, that I'm saving for a project like this. The transmitter made 300W from four TV sweep tubes, and the PT has about a half dozen different HT windings and about 12A of heater current.

You lucky bastard.