Ultra linear and phase inverter drive

[Tage]

So I'm building a Bass amp and looking to run about 150-200 watts out of it. I have most of my bases covered but I'm wondering if I should use the ultra-linear taps on the OT, I'm using the Weber one rated for 300 watts with a 1.9k primary, with 4 6550's. Now my customer realy likes bass amps like the orange AD200B and the old Highwatt 200 and 400 watt amps, which are amazing amps to see. So I'm thinking the power amp distorting is gonna be expected, but if I recall correctly Ultra-linear amps don't have the nicest power amp distortion, or is that untrue?

Also I know many amps ake this kinda power with a normal LTP PI, using a 12AT7 to drive the power tubes as in the Highwatts. But I was wondering what the benefit is to have cathode followers on either side of the phase inverter, kinda like in the SVT, other than more current to drive the grids? Does this method increase output signal swing?

[exclamationmark]

You can always make the screen grids UL/resistor switchable to see the difference for yourself. I haven't actually played any UL amps myself, so I wouldn't know about the breakup characteristics. Although... since it is a type of NFB, the distortion may have an abrupt sort of quality to it.

Depending on your cathode followers, it can have a significant impact on how much power it can output. When the grids of tubes approach 0 volts, the resistance of the grid drops from some arbitrarily high value to several 1000 ohms. Given enough current, you can push the grids of the tube positive and it will keep amplifying as if nothing has happened. Some big power tubes are even designed for their bias to sit at 0v, and amplify with their grids going both positive and negative in respect to the cathode. When the grid is pushed positive and is actually amplifying it's called class AB2 operation.

So a cathode follower that can source lots of current can potentially push the grid of the power tube largely positive, and increase output power. Depending on the power tubes, this increase can be substantial. I could get 150% more current from a 6as7g by pushing it's grid to something like +9v, and this particular tube is actually designed for grid current operation with radiators attached to the grid. As for a conventional 6550 stage, you could do a search for class AB2 operation.

Ever since I found out about mosfets, I just use a cathodyne PI coupled to a couple of mosfet source followers, which are DC coupled to the power tubes. Since the cathodyne doesn't have to source much current, I avoid the nipple distortion effect, I save a few resistors and I save a whole gain stage!

[Gingertube]

Ultralinear operation reduces the output tube internal impedance (rp) to almost triode connection values. Distortion will be less and bass will be tighter. I personally like Ultralinear for Bass Amps. Many others don't.
Cheers,
Ian

[Gtr_tech]

UL operation tends to suit bass amps well.....as long as you're not looking for breakup. They tend to sound a bit ugly when pushed into clipping. But a "side effect" of UL operation is an upper mid hump that works well for a bit more clarity. That clarity in a guitar situation translates to screechy, so its not so hot for a nice overdriven guitar sound (via pedals).

Pedal steel guys like the UL stuff as well because of this tonal shift toward upper mids...

[trobbins]

For a cleaner sound, you should probably aim to make the B+ quite stiff, with lots of capacitance (eg. 1mF) and SS diodes. You should probably aim to limit/saturate the signal further back in the chain, possibly before the PI, and not attempt to get in to hard saturation of the output stage itself.

[Chuck H]

So I'm thinking the power amp distorting is gonna be expected, but if I recall correctly Ultra-linear amps don't have the nicest power amp distortion, or is that untrue?

I think Dr.Z would take issue with that.

Also I know many amps ake this kinda power with a normal LTP PI, using a 12AT7 to drive the power tubes as in the Highwatts. But I was wondering what the benefit is to have cathode followers on either side of the phase inverter, kinda like in the SVT, other than more current to drive the grids?

I think the point to the higher current grid drive in some bass amps is to achieve operation closer to AB2. In this case you really DON'T want to clip the power tubes because of the likelyhood of excess crossover distortion. But, using cathode followers to reduce impedance, and therefor grid loading IS an effective way to REDUCE crossover distortion in AB1 operation and therefor improve the nature of the clipping. Very generally speaking, AB2 would be a cooler bias and more grid swing. But the tubes are "on" for a shorter time than AB1, but conducting more than AB1. This condition of such high dissapation requires that the grid can draw current. AB2 can make more clean power than AB1. But AB2 is prone to crossover distortion when clipping because just as the tube spends less time "on", it spends more time "off". This longer off time is what keeps the tube from overdissapating while creating more power. But shows up as crossover distortion when the tube saturates. AB1 would be a hotter bias and lower grid swing. That is, the tube spends less time in cutoff and because the tube is already idling closer to saturation it requires less grid swing to saturate. Having a shorter off time keeps the pesky crossover distortion at bay but also lowers available output power. In AB1 there is no grid current. Ideally. There is an effect called grid loading that, combined with the longer time constant of the typical high impedance cap coupled PI, causes a charge to build on the PI coupling caps because the caps don't have time to fully discharge. This charge appears to the tube as a second bias voltage and cools the bias, causing crossover distortion due to the increased off time for the tube. The more you drive the grids, the more they load. So this phenomenon usually manifests along with saturation. Using a PI capable of some current usually means a lower impedance as well. This lower impedance also has a lower time constant. This allows the PI coupling caps time to discharge and reduces grid loading and therefor crossover distortion in AB1.

Whew... It's worth noting that, for AB1, since current isn't really the issue, the time constant is, that any method of discharging the caps is an equally suitable solution. There is a nifty circuit called the Paul Ruby mod that uses diodes to allow a discharge gate for the PI coupling caps that is only open while the power tube is is in cutoff. So there is no audible effect from the diodes other than the reduction in crossover distortion because the grids can discharge.

If some power amp clipping will be expected I would use AB1 and skip the cathode followers, but use the Paul Ruby mod instead.

Does this method increase output signal swing?

No. Just PI output current availability.

EDIT: This is a laymans explaination. Which is the only explaination I have. But I thought the perspective was relevant. Any technical niggles accepted.

[Tage]

Can you expand on those mosfet followers? it would require me having a lower power supply but might be worth while.

[Tage]

Thanks Chuck, laymans terms is what I'm looking for. I do believe the power amp will be pushed into distortion regularly and that is the kind of bass tone my customer is wanting, so Class AB1 operation is what I'll go for, most likely an 12at7 with lower load resistors around 33-47k, as i'll still get good gain and good signal swing for both overdrive and full, or close enough to full power from the valves. I've used a calculation to test and see what kind of values would help reduce that "bias excursion" and it seems to me that smaller coupling caps are the best way to go, although I'd like a way to keep them big for lots of low end, though not too low. You wouldn't happen to have a schematic of this Paul Ruby mod do you?

Oh and I had heard that even some trainwrecks used UL output sections, so they cannot be all that ugly sounding. My only experience is with a few Silverface UL amps, and they did shift really abruptly into distortion, and that amp belongs to this customer and he is not a fan. I suppose I should have written off the UL thing based on that, but I really like the input of this forums members

[exclamationmark]

Can you expand on those mosfet followers? it would require me having a lower power supply but might be worth while.

Search for MOSFET follies by R.G Keen. Realistically, unless you design the amplifier with class AB2 in mind, the extra output power won't be that noticeable. This entails a somewhat lower voltage, in order to allow your class B load line to have a steep gradient to take advantage of extended grid curves. If you take a typical output stage and slap a driver circuit to push it into AB2, you'll literally get perhaps 10 or 20 volts of extra swing and maybe 10-20 mA of extra current. Not much. You really want to load line to pass above the knee of the 0v grid curve.

[Chuck H]

The Paul Ruby mod is explained ad nausium in several threads here. I'm suggesting a search since re-explaining it seems redundant. The principals are simple and the results for AB1 are very satisfactory. Just to reitterate... In general, AB2 gives more clean watts but poor overdrive. AB1 doesn't require that the PI can supply current, but rather only needs a discharge opportunity for the PI coupling caps. The Paul Ruby mod can give you this without the need for low impedance systems like cathode followers or MOSFETs.

[Steve Conner]

The "nipple distortion effect" was mentioned above. WTF is that?!

I ask because in another thread, someone posted a scope screenshot of a clipped wave with a nipple-like appearance to the peaks on one side. I'd never seen anything like it before, although I have seen the peaks fold inside out when overdriving beam tetrodes with too high load impedance. (maybe that is cleavage distortion? )

[exclamationmark]

The "nipple distortion effect" was mentioned above. WTF is that?!

I ask because in another thread, someone posted a scope screenshot of a clipped wave with a nipple-like appearance to the peaks on one side. I'd never seen anything like it before, although I have seen the peaks fold inside out when overdriving beam tetrodes with too high load impedance. (maybe that is cleavage distortion? )

I actually have no idea how it occurs, but according to merlin it's caused when the output tube on the cathode of the cathodyne begins to draw grid current. Normally it would have unity gain, but because of the onset of grid conduction from the output tube, the cathode output is clamped, which means the negative feedback afforded by the unbypassed cathode resistor is no more (due to a constant cathode voltage - essentially the same as fixed bias). This causes the gain to suddenly increase, which is reflected at the anode in the form of a spike (although we all know it's really a nipple ).

[trobbins]

The Valve Wizard

[Tage]

Yeah I did a search for it and I believe I will implement that in this amp, thanks for the heads up.

[loudthud]

The "nipple distortion effect" was mentioned above. WTF is that?!

I ask because in another thread, someone posted a scope screenshot of a clipped wave with a nipple-like appearance to the peaks on one side. I'd never seen anything like it before, although I have seen the peaks fold inside out when overdriving beam tetrodes with too high load impedance. (maybe that is cleavage distortion? )

I've seen this distortion on the grids of 6550s in an amp I was prototyping. It was caused by clipping on the 6550 plate being fed back to the phase inverter which "slams the rail" trying to increase the output beyond clipping. In this amp, the peak grid voltage when the output clipped was about -10V. The nipple would grow up to 0V where grid current would begin. The open loop gain was so high, the nipple looked more like a top hat, nice and square.