Cathode Follower with Plate Resistor?

[GregS]

I'm currently trying to design a bass amp with a line out. I was hoping to use the second gain stage of the pre-amp as both a regular gain stage to drive the phase inverter and as a cathode follower to provide a low impedance signal for the line out. So I'm trying to do two things in one amplification stage.

I can't seem to find an example in any of my books where this is done (i.e. all example cathode followers have no plate resistor) so I suspect that this isn't normal practice for a reason. One thing I've already observed from scoping an existing amp (regular triode gain stage with plate resistor) is that the signal at the cathode is considerably less than that input to the stage at the grid - significantly less than unity gain. I've figured this to be the effect of the plate resistor.

Another thing that occurred to me is that the signal at the line out will be out of phase with the signal at the speaker if I'm not mistaken. I don't think this would be desirable in a gig situation where what's coming out of the PA is out of phase with what's coming out of the speaker cabinet on stage. I would think the lower frequencies would cancel.

Thoughts on this?

If this is a bad idea for the reasons I've mentioned or other reasons I haven't thought of then I'll probably just scrap the line out idea altogether. I don't want to go to too much trouble to implement one as simply mic-ing the cab will probably get better results than any line out anyway.

Any input would be much appreciated.

Thanks,
Greg

[defaced]

It sounds like you're describing a sort of cathodyne phase inverter circuit, which on it's own has slightly less than unity gain. The Valve Wizard Since cathodyne's have more or less equal plate and cathode resistors, there may be some room to fudge either the voltage gain or the output impedance of the follower to get more/less of one characteristic - at least with my basic understanding of what's going on this makes sense.

As an alternative, I'd look to MOSFETs to run the line out. Take an LND150 and set it up for unity (or whatever gain addition/reduction you need) to invert the phase so the line out and speakers are in phase and use an IRF820 as a source follower to buffer the output. If set up properly, both will be full bandwidth and dead clean. R.G.s MOSFET Follies will cover the IRF820 stuff, and the LND150 is a depletion mode device, so it biases up like a typical tube triode.

[soundmasterg]

It sounds like you're describing a sort of cathodyne phase inverter circuit, which on it's own has slightly less than unity gain. The Valve Wizard Since cathodyne's have more or less equal plate and cathode resistors, there may be some room to fudge either the voltage gain or the output impedance of the follower to get more/less of one characteristic - at least with my basic understanding of what's going on this makes sense.

As an alternative, I'd look to MOSFETs to run the line out. Take an LND150 and set it up for unity (or whatever gain addition/reduction you need) to invert the phase so the line out and speakers are in phase and use an IRF820 as a source follower to buffer the output. If set up properly, both will be full bandwidth and dead clean. R.G.s MOSFET Follies will cover the IRF820 stuff, and the LND150 is a depletion mode device, so it biases up like a typical tube triode.

I'd also suggest the MOSFET route myself. A cathode follower doesn't have any gain and is there to buffer things on the other side from a gain stage in front of it. As soon as you add a plate resistor then you have gain, and you progressively lose the buffering effect of the CF. You could add a gain stage and a CF, but in situations like this, SS works better once you know how to use it. You are already most likely using SS pedals with your proposed loop too so what difference do a couple MOSFETS make?

Greg

[Enzo]

A cathode follower has less than unity gain, but not tons less. WHen you are measuring this signal, is there a bypass cap on the cathode in question?

[Gingertube]

If you pick a point in the circuit with the appropriate signal level AND you do not use a bypass cap on the cathode resistor it is perfectly OK to take the amplified feed from the anode to drive the next stage of the amp and to take the line out feed from the low impedance (un amplified by this stage) cathode. I've seen this done in a commercial amp and it was discussed on the Powerscaling forum but I could'nt track it down with a quick search.
I would go the mosfet source follower route myself. The ZVN0545A 600V rated mosfet is recommended. It is a little E-Line device 600mW rated.
Need to know more about MOSFET Source Followers? Do a websearch on "MOSFET Follies". There are some "standard" circuits shown.
Cheers,
Ian

[Chuck H]

When you consider that a "line out" is only about a volt (actually .9???) it wouldn't be hard to get that from the cathode of a standard gain stage. I might start with a 5k pot (wired as a variable resistor) for a cathode resistor (unbypassed) and decouple via a cap to an output jack and then a 22k load resistor. Adjust the pot to get about a volt on the jack at fairly high volume settings. Now take out the pot and measure the resistance so you can replace it with a fixed resistor of that value. This method allows you to use this triode as a gain stage too. Though limited in output swing it'll still beat the pants off a cathode follower or cathodyne for driving the PI from the plate. The impedance of the circuit should be low enough to handle cables and not cause problems with effects processors.

[Enzo]

But not if there is a bypass cap.

[GregS]

Thanks for the replies.

WHen you are measuring this signal, is there a bypass cap on the cathode in question?

There is. It is a low value cap to provide a high frequency boost. Before testing it I thought that only the very high frequency AC signals would be effected. However, when I sweeped through the frequencies on my signal generator I found that the signal started dropping off just above 100 Hz and continued to drop steadily with increasing frequency. So, I could not use a cathode bypass cap here. At the lower frequencies I found the signal at the cathode was roughly half that at the grid. I also tested the signals at the grid and cathode of the first stage, which has an un-bypassed cathode resistor, and also found the cathode signal to be roughly half that at the grid. Both stages use the same value plate and cathode resistors and the same HT supply.

I'll look into the MOSFET solution. It sounds interesting and it would be good to learn this.

Another idea came to me to simply use another dual triode tube for the line out path. The signal level input to the second stage is too small to get line level off the cathode even if I got full unity gain so I need another amplification stage before the cathode follower anyway. Also the the amplification stage would invert the signal so that the signal at the line out and speaker would now be in phase. A voltage divider before the cathode follower stage would set the input signal to ideal level for a line level signal at the cathode. This solution also allows the second gain stage to be uncompromised by the line out - so a cathode bypass cap can still be used and the plate and cathode resistor values can remain as is also.

I'm assuming that to implement either the MOSFET or the dual triode tube solution the grid of the second stage would also be connected to either the grid of the new tube or the input of the MOSFET? The circuit paths would branch here and be completely independent from each other afterwards.

Also, the line out is intended to go directly to a powered mixer for the PA. Not to any effects or other power amps. I read something about the need to remove some of the very high frequency content to get a more accurate simulation of a mic-ed speaker cabinet if used this way (direct to PA). However, this may have been referring to a guitar application. Perhaps you would not want to limit high frequency content for bass? I could put in a low pass filter at the line out to accomplish this. Thoughts?

Thanks,
Greg

[jmaf]

The bypass cap may be small for its function, but the coupling cap right after it is normally even lower in value, in which case the lowest impedance to ground is via the cathode bypass.

My 2 cents: if you draw your planned circuit it's easier to discuss it, also while drawing you'll arrive at conclusions about it, it's like stating a problem out loud instead of thinking it out, lots of folks solve the problem in the middle of explaining it to someone else.

[imaradiostar]

If it's a line out why not use a transformer from a higher voltage point in the circuit? Do you intend to have a corresponding return? If not just tap the speaker output with a resistive divider and a transformer.

Maybe I'm missing the point.

jamie

[GregS]

My 2 cents: if you draw your planned circuit it's easier to discuss it

You're right. A picture is worth a thousand words. I'll report back with some scanned in drawing(s). It will be at least a few days before I can get to this.

If it's a line out why not use a transformer from a higher voltage point in the circuit? Do you intend to have a corresponding return? If not just tap the speaker output with a resistive divider and a transformer.

Jamie, that's an interesting suggestion. However, I don't want to take the line out signal from anywhere in the power amp or at the speaker as the line out signal level will be affected by the volume control (which will be in between the output of the pre-amp and the input of the phase inverter.

Thanks,
Greg

[imaradiostar]

You're overthinking the phase relationships for the bass. You have no control over how many times the phase inverts between your amp and the subs in the house pa. By the same token a decent sound guy will be able to compensate for the level change if your line out level changes slightly because you bumped up your volume.

If you're going to a house PA and not using a direct box you're going to almost universally want a transformer to isolate you from shocks and ground loops and prevent external loads from altering the tone of your amp. I was playing my SWR 350 for a church Christmas concert and the PA was making terrible noises- because the XLR out wasn't transformer or capacitor isolated and when they turned on global phantom power the Mackie console and my amp were in a battle- in the end the console won because my amp wouldn't function at all with 48 volts across the line output terminals! I haven't yet had a chance but I will be installing an isolation transformer which will render this a non-issue.

If you're building this amp to achieve a certain amount of sonic coloration the DI transformer being fed from the a resistive divider at the output makes the most sense as it'll give you an output with all the coloration the amp provides.

If a dead clean line out is what you desire the mosfet source follower driven from the plate of your first gain stage makes the most sense. I would still tend toward a transformer after the mosfet for isolation. Failing that make sure your line out is being plugged into a decent DI. You could drive the source follower from any other intermediate stage depending on whether or not you want the amp's EQ, gain controls, and inherent coloration to effect the line out.

jamie

[GregS]

You're overthinking the phase relationships for the bass. You have no control over how many times the phase inverts between your amp and the subs in the house pa. By the same token a decent sound guy will be able to compensate for the level change if your line out level changes slightly because you bumped up your volume.

If you're going to a house PA and not using a direct box you're going to almost universally want a transformer to isolate you from shocks and ground loops and prevent external loads from altering the tone of your amp. I was playing my SWR 350 for a church Christmas concert and the PA was making terrible noises- because the XLR out wasn't transformer or capacitor isolated and when they turned on global phantom power the Mackie console and my amp were in a battle- in the end the console won because my amp wouldn't function at all with 48 volts across the line output terminals! I haven't yet had a chance but I will be installing an isolation transformer which will render this a non-issue.

If you're building this amp to achieve a certain amount of sonic coloration the DI transformer being fed from the a resistive divider at the output makes the most sense as it'll give you an output with all the coloration the amp provides.

If a dead clean line out is what you desire the mosfet source follower driven from the plate of your first gain stage makes the most sense. I would still tend toward a transformer after the mosfet for isolation. Failing that make sure your line out is being plugged into a decent DI. You could drive the source follower from any other intermediate stage depending on whether or not you want the amp's EQ, gain controls, and inherent coloration to effect the line out.

jamie

Thanks Jamie, lots of good information in your post. There seems to be a lot of potential problems associated with line outs in general if an isolation transformer isn't used. In most situations we use our own PA and soundman so have control over what's going on there. There was an occasion though where the line out of my present amp couldn't be used as there was massive hum - ground loop? I had to use the direct box for that.

My idea is to have colouration in the line out signal and I see that taking it at the output of the power amp will have the colouration of the entire amp. I'm a little confused about how to implement a resistive divider here though as, the way I see it, it would load the output of the amp more. I imagine the resistors would require a very substantial power rating as well as this will be a high power amp.

You've got me thinking. A line out without an isolation transformer is not a perfect solution as you've pointed out. With the isolation transformer things get a little more involved than I had wanted.

Like I said in my first post, a line out isn't a must. It's more of a "nice to have" so just using the direct box is also an option even though there's no sonic colouration going to the PA with the direct box. Even if decide to drop the line out for this amp I will have to revisit all of this again in the future as I'm planning a recording pre-amp project. I'm sure isolation transformers will be necessary for this.

Thanks for your advice. Much appreciated.

Greg

[imaradiostar]

No problem. It always helps to understand where you're headed and the reason behind your choices.

Check out Edcor transformers- they're solidly made and very cheap. I've used them for scientific equipment at work with excellent (dead clean) results. You can get a small PC mount type Edcor that can be mounted on a PC board or eyelet board and you can experiment with connecting it to different points in your circuit. You could even put the mosfet source follower, some minor filtering and bias circuit and an "input level" pot all on one board and you'll basically have made an active DI to connect to any random point inside your amp.

If you're connecting a DI transformer across an audio output you don't need any particularly large resistors. Remember the basics:

Volts=amps*resistance
Watts=amps*volts

From those two formulas you can derive all kinds of things.

Let's do an example case.

Say you have a 4 ohm bass cab. Assume you're playing hard and putting 100 watts into the cabinet.

volts/resistance=amps so

watts=(volts)(volts/resistance)

4*100=v^2

volts= (sqrt400)

volts=20

So from that we know if your bass amp is cranking out 100 watts into 4 ohms it'll be 20 volts rms.

Let's assume you're using a 1:1 transformer. If you want to output a reasonable level for a console with a balanced input I'd say 1 volt RMS is a good number to shoot for. Most modern consoles have an input impedance that's high enough that I'd call it "negligible." In other words it's not a deal breaker if you're not driving the transformer from a perfect 150 ohm load.

So...I'd take a 2.2k resistor and connect it to the speaker terminal output internally and one end of the transformer primary winding. I'd put a 100 ohm resistor across the primary winding. Other end of the primary & resistor go to circuit ground.

this would give a level reduction of:

100/(2200+100)=.0435

Which means your 20 volt amp output would now be .87 volts into (and out of) the 1:1 transformer. A comfy level for a console that should be well above the average mic.

The resistors will dissipate only the voltage across their impedance. Rearranging formulas again we get:

w=v(v/r)

20(20/2200)=w

.18=watts

That's for the 2200 ohm resistor. The 100 ohm would be far less. You can use any normal 1 watt resistor with plenty of safety margin (up to around 400 watts output at 4 ohms!)

I'd be tempted to use a 1k resistor feeding a 1k pot with the wiper tied to the transformer.

OK, gotta get back to Java programming. This being an adult in school stuff sucks.

jamie

[GregS]

Thanks for taking the time and effort to explain this Jamie. I like the sounds of this and I think I'll try it out. I had been concerned about the size of the additional transformer and possibly big power resistors but if the transformer is small enough to mount internally on a circuit board, is cheap, and regular 1W resistors can be used then this seems physically easy to implement.

Also, as I understand, it will not affect the load the power amp sees as the speaker load parallel to these resistors is essentially the same as the speaker load itself (i.e. 8R || 2300R is essentially 8R).

The other benefit of this solution is you get the entire amp, minus the speaker of course, at the line out instead of just the pre-amp as you pointed out earlier.

I'd be tempted to use a 1k resistor feeding a 1k pot with the wiper tied to the transformer.

As I understand, this allows one to compensate for the volume the amp is set at so the level of the line out is not entirely dependant on the amp's volume. This is a good feature!

OK, gotta get back to Java programming.

Actually, I'm a programmer as well. An IBM Mainframe old school type. I have been doing some work with more modern object oriented stuff like .NET/C# in recent years though.

Thanks again. Much appreciated.

Greg