Ultra linear and phase inverter drive

[Alan0354]

I am not an expert in power amp, in fact I am still learning.
My understanding of an op-amp is the closed loop feedback linearized the amplifier. In order to get more linearity, you raise the loop gain ( log subtract open loop gain and feedback gain). The more loop gain you have, the better it compensate the distortion and you get less distortion.

Most of the tube power stage has a feedback resistor from the secondary of the transformer back to the differential stage just like an op-amp. So the key is to raise the forward gain of the power amp to increase the loop gain overhead therefore improve linearity. Yes putting a cathode follow buffer the differential stage and the gain will be higher. Or even add another stage. But this will make closed loop pole and zero compensation harder. This together with using the linear tape of the output transformer should minimize the distortion. An over rated OT will help also. Look at how big the Fender Bassman 100 OT. I made a mistake of modifying a BM100 for distortion amp, that thing just don't sound right no matter how hard I drive.

Please excuse me as I am a beginner in this kind of power amp, but the theory should be sounded.

[imaradiostar]

If you use a DC coupled source or cathode follower and all the supplies are fully bypassed you're not adding another pole.

What do you mean about the bassman 100 output? Are you saying it's too big or too small?

jamie

[Tage]

Most of the write threads I have found regarding the Paul Ruby mod are all addressing cathode biased output sages. I was most likely going to go with fixed bias, in which case i will need Zeners or a chain of Zeners that can cover a range of up to -75V of fixed bias voltage, although I'm sure my actual bias voltage will end up being some what lower than that. Am I missing something here? Chuck?

[imaradiostar]

Smarter guys than me can speak to it but I think it's a mistake to confused the ruby mod with using cathode or source followers to drive the output grids. The sonic result may be similar but I have a fair suspicion (and Ken Gilbert will likely chime in here) that driving the grids from a cathode follower will make a big difference in a bass amp.

That said- the ruby mod is the perfect fix for a small cathode biased amp!

jamie

[Alan0354]

If you use a DC coupled source or cathode follower and all the supplies are fully bypassed you're not adding another pole.

What do you mean about the bassman 100 output? Are you saying it's too big or too small?

jamie

I meant Bassman 100 output transformer, just missing a word. The transformer is too big and linear for guitar amp to get any compression.

Poles and zeros are not about DC coupling. If you look at the Bode Plot of op-amp. you will see an open loop gain plot which in this case is the forward gain of the power amp from the differential stage, through the power tube and through the output transformer. For example the output transformer has inductance where the output voltage start to decrease at some frequency, that is a pole and the pole freq is the freq where the output is down to 0.707 of the amplitude at low frequency even if the input is constant. Also the 100K plate resistor( just an example value) together with the capacitance of the grid it is driving form a pole where the amplitude start to roll off at some higher frequency. This is all in the forward gain path, each poll will cause a 90 deg phase delay( phase lag) and the total phase shift eventually become 180 deg due to the two poles. The op-amp feedback theory is to feed the signal back to the negative input, so it is 180 deg off. If you add the extra 180 deg from the two poles in, it would be 360 deg. That become possitive feedback. If the amplitude gain is still greater than 1, the amp will burst into oscillation. A lot of times people put a small cap in parallel to the feedback resistor that connect from the secondary of the OT to the differential stage to create a zero to neutralize one of the two poles as described above to stabilize the amp.

That said, increase the forward will bring the graph of the open loop gain up and if the totally phase shift is 360 deg, you have a problem. There are different ways to tame this kind of pole zero stuff and is well documented. Point is to raise the forward gain, tame the poles and you should have a more linear amplifier even if you don't do anything else...............At least it's the theory. I am not saying this is the best solution as I am not an expert, I just throw this in as another point of view.

But the question is are you sure that's what sound the best. People intentionally get rid of the closed loop feedback of the power amp to get better sound due to distortion. As all you experts in amps here know there are amplifiers that don't have feedback.

[Tage]

Well Alan I'm not planning on using too much Negative feedback in this build as I believe that will help with generating a nice overdriven bass tone. Also I have read up on ensuring my power tubes are my dominant pole and all that stuff Merlin's wonderful books have taught me, now he just needs to write his power amp book and I would be asking a whole lot less questions.

Anyways, there will be a lot of testing and adjusting to get the best response and to ensure this amps character is what my customer wants. the Paul Ruby idea with the Zeners preventing/clamping the negative voltage swing seemed like a good idea to me, I just am having trouble implementing it in a 180 watt amp...

[imaradiostar]

I see what you're getting at. I would argue that while the science you use is sound, the application of it at musical instrument frequencies is somewhat less clear.

Assume you have a common Fender style output- cathodyne or long tailed phase inverter, it doesn't matter. You'll get a phase shift in a few places- coupling cap at the input grid, coupling caps between the PI plates and power tube grids and through the output transformer. The poles of the plate impedance (37ish k, less with a 12au7) and the picofarads of a tube's grid are well out of the audible region and a correctly designed musical instrument amp usually has those frequencies rolled off anyway.

The above circuit, set up correctly, uses negative feedback and is quite stable.

Adding a mosfet or triode with the source or cathode directly coupled (maybe through a small value grid stopper) to the output valve's grid will not add a pole- it will only serve to eliminate blocking distortion and, if set up appropriately, drive the output grids slightly positive relative to their cathodes.

As for output transformer sound, what you're saying flies in the face of many designs with proven tonal qualities and oversized iron (certain marshall, trainwreck, komet, etc). I would argue that other factors would prevent that amp from being a good "distortion" amp. I would also argue that an undersized output can have interesting tonalities but it's hardly the difference between "good" and "bad" distortion tone. A too-small output tends to manifest itself in the bottom octave of the guitar first and sounds largely the same above that.

We must, of course, keep in mind that "good" sounding distortion is such a subjective thing!

Good conversation here. I hope what I'm saying doesn't ruffle feathers!

Jamie

[imaradiostar]

Anyways, there will be a lot of testing and adjusting to get the best response and to ensure this amps character is what my customer wants. the Paul Ruby idea with the Zeners preventing/clamping the negative voltage swing seemed like a good idea to me, I just am having trouble implementing it in a 180 watt amp...

I had another thought- hope it doesn't muddy the water too much!

The more important thing, seeing as it's a customer amp and not a personal amp, is reliability! Reliability with modern big bottles requires smallish grid leak resistors. Personally I'd shoot for 47k to 100k but that's only because I've spent some time torturing some tubes and landed at those values. Smaller is generally better, especially if they're run hot.

Unless you intend to use a very low Ra valve for the phase inverter you'll have a hard time driving 24k or 50k per phase (or less) effectively. Using mosfet source followers allows individual bias adjust per output tube, DC coupling to as low a source R as you choose, and the ability to drive the grids slightly positive, something that could be helpful if you don't have 700+ plate volts to play with.

Just thinking out loud.

jamie

[Tage]

Yeah for my initial design scetch I had 68K grid leaks, and this is why I wanted to use a 12at7 phase inverter set up, as it seems to be a good balance of current drive yet still with decent gain. I know it's not the greatest sounding PI when overdriven, but really I wanted to use it to overdrive the power amp. If I will still benefit from source followers or cathode followers after the PI then I may implement them, but I think I should be ok to drive the 6550's to full output with only the 12at7 LTP.

[imaradiostar]

I think you're right, a 12AT7 would probably do it. I'd be tempted to just cut to the chase and use a 12AU7 or even a 12BH7 and just drive the phase inverter harder!

I have a not-too-dissimilar project in the works with an Edcor 100 watt transformer which the designer assures me should do a solid 150 watts at 40hz, if not closer to 200. I'm mismatching a 6.6k output down to 3.3k so I'll get 2, 4 and 8 ohm taps out of the deal. I'm undecided if it'll be 4 6550's or 2 KT120's. I'm leaning toward the 120's as I have a PT that should be a good fit to make around 150 watts. That's more than enough for my bass needs with a 4x10.

(this is where Ken Gilbert steps in and says "sure it'll work, but it won't sound as good as driving from the cathode!")

I'm starting to think a big clean tube bass amp with minimal coloration and basic EQ is the way to go. More of the bass battle is fought and won with a pedal or two (tech21 VT bass/big muff) and a really efficient cabinet (bergantino NV610 or SWR Goliath SR).

I'm excited to see how yours turns out. I had planned to build mine over winter break but like so many things in my life...not enough time!

Jamie

[Alan0354]

Absolutely not, I just throw this out so we can talk. We all know all talks are cheap, you can talk theory until your face turn blue and only thing matter is how the customer like the sound. The op-amp theory is sound, but whether this is desirable is a totally different question.

I am just guessing the size of the transformer. I have no idea how this relate to the real world sound. Also I based on looking at the size of the transformer used in Mesa. They either are real cheap or they also buy into the some core saturation faster and compress the signal. Yes, I agree that the Marshall OT are quite big for the rated wattage compare to other brand and their distortion is second to none in my book.

I don't have a lot of experience on the power amp, just join in the conversation. Still have a lot to learn.

[Chuck H]

Smarter guys than me can speak to it but I think it's a mistake to confused the ruby mod with using cathode or source followers to drive the output grids. The sonic result may be similar but I have a fair suspicion (and Ken Gilbert will likely chime in here) that driving the grids from a cathode follower will make a big difference in a bass amp.

Yes, using a low impedance to drive the grids of a bass amp will give maximum clean headroom, but, this applies mainly to a bias condition that requires grid current. In that case the amp will be in, or close to AB2. Since the OP intends to offer a build that can clip the power tubes in a musical way the amp should probably be in AB1. In which case devices allowing for grid current don't offer any advantage. As I mentioned before, The use of low impedance grid circuits in AB1 amps doesn't offer any advantage because it can supply current. The advantage is the lower time constant. This allows the PI coupling caps to discharge. The additional current availability isn't needed for AB1. Any circuit that allows the PI coupling caps to discharge is equal to the task. Making the Paul Ruby mod a better choice because it simplifies the circuit, reducing phase errors, and is easier and more cost effective to implement.

That said- the ruby mod is the perfect fix for a small cathode biased amp!

Or any AB1 amp where you might want to reduce grid loading.

[Chuck H]

As for output transformer sound, what you're saying flies in the face of many designs with proven tonal qualities and oversized iron (certain marshall, trainwreck, komet, etc). I would argue that other factors would prevent that amp from being a good "distortion" amp. I would also argue that an undersized output can have interesting tonalities but it's hardly the difference between "good" and "bad" distortion tone. A too-small output tends to manifest itself in the bottom octave of the guitar first and sounds largely the same above that.

We must, of course, keep in mind that "good" sounding distortion is such a subjective thing!

Good conversation here. I hope what I'm saying doesn't ruffle feathers!

Jamie

+1
A bass amp needs to make bass frequencies. Small iron pretty much sucks for that. The tubes will overdrive in a useful way with proper working conditions. No need to try and psuedo EQ with undersized iron. In fact it IS the rolloff of bottom end that is most often useful about an undersized OT. Useful for keeping a six string guitar from sounding boomy. "I" prefere that my bass amps have plenty of boom.

[imaradiostar]

I think Chuck is pretty much right. I just don't think it's quite as simple as "this one is better" and I don't know that bias point has as much to do with it as other aspects.

I think there may have been a disagreement about this before- the real advantage, per KG, was the complete lack of cap between the output grid and the element before it- this way there is no time constant and no grid blocking at all. By having sufficiently high voltages above and below ground for the cathode or source follower it remains completely linear (ie, no grid blocking there either) even while the output tubes are starting to draw grid current. This advantage remains regardless of operating class. This, I would argue, is plainly audible and makes a big difference in any big power amp.

You're right to point out the time constant. Unfortunately the lower impedance does nothing to help- now you have a source that can charge the caps more quickly. In the end the only way around this is to reduce the coupling cap values so they discharge quickly and use the zener mod to force them to discharge or not even get charged in the first place.

While I'm pointing out that they largely achieve the same thing there are differences that are crucial, especially as it applies to how the amplifier is used.

Be mentally prepared to hear grid blocking at the outputs if you're going for "output distortion" with big bottles and capacitively coupled bias- then probably after a while you won't hear anything because big bottles are expensive so you'll be broke (and probably deaf from the extreme volume.)

It's one thing to add a little crunch like an SVT, it's another thing entirely to push big bottles like you would the EL84's in an 18w marshall circuit. The zener mod is a great way to mitigate the "zzzzzzzz." Large value grid stoppers can really help too. If you've got low enough driver impedance from a LTPI it's worth looking up the input capacitance of your output tubes and sizing the grid stopper for a rolloff just above the highest frequencies you care to reproduce. That value might be surprisingly high. You'd need to lower your bias grid leak resistor accordingly.

Dangit I need to go to sleep. This is just too darn interesting.

jamie

[Chuck H]

Most of the write threads I have found regarding the Paul Ruby mod are all addressing cathode biased output sages. I was most likely going to go with fixed bias, in which case i will need Zeners or a chain of Zeners that can cover a range of up to -75V of fixed bias voltage, although I'm sure my actual bias voltage will end up being some what lower than that. Am I missing something here? Chuck?

The rub with the PR mod is that to get maximum benefit you'll want the ZV only a small amount higher than the actual bias V. So, when rebiasing you might find you also need to change the ZV of the diodes (or diode string). But you can still get plenty of benefit with the ZV set suitably high so as to likely never be below any reasonable bias voltage. 75V is almost certainly higher than you need to go. Diodes are cheap. So get a few extras for tuning. If you find, for example, that you have a bias voltage of -55V, and you figure any reasonable bias for any tube probably won't need more than -65V bias voltage, set the PR mod ZV at 66ZV. Whenever the PI swings beyond 66V on the bottom of the waveform the power tubes will certainly be in cutoff and the PI coupling caps will be discharging. Not quite as effective as using a ZV of 56 for a -55 bias, but still better than nothing. That said...

The low impedance drive suggested by others here would offer the same benefit at any bias setting. So...

Looks like things are either more finicky OR less beneficial with the PR mod. Or you can go through the finicky process of implementing low impedance grid drives and never have to deal with the limitations again.