Thor mu amp stage/section

[dthurstan]

Hey

http://www.runoffgroove.com/thor.html

I notice the thor has a mu amp stage that "simulates the.. Class AB push-pull" stage. I'd like to add this to my own emulation. I'm not sure how to adapt this to my own project.

I'm looking at building bassman jfet emulation. The finished project will be based on 59 Bassman (using MartyMart's BassBoy 59 schematic as a starting point) with 3 band eq, presence (probably as a further tone control I have seen someone incorporate this into a Marshall emu on this forum). I'm also including a cab sim (ROG Condor). So I'll have an o/p from the emulation to go to a guitar amp or an o/p from the cab for direct recording. Also an i/p into the cab for maximum flexibility  .

Is this section emulating the power stage of a Marshall (and therefore adaptable to represent a Bassman power stage) or is it more a tonal stage adding odd order harmonics as a power stage would?

BTW I realise that this emulation will in no way give a bassman in a box  I'd just like to incorporate as many stages as possible (pre amp, eq, power amp, cab).

Cheers

Dave

[Davelectro]

Mu amps are really great.

Don't forget to emulate the response of a transformer-coupled output stage. Check out Marshall's FDD patent, for example. It makes a big difference, IMO.

[sault]

Well, in this case technically it's an SRPP stage. You can see some basic Geofex notes here that might help...
http://www.geofex.com/Article_Folders/modmuamp/modmuamp.htm

Anyways, yes, it doesn't have much (if any) 2nd order distortion, so its a way to mimic the "push-pull" sound. While I vaguely remember that it loses gain if it isn't loaded down, I don't remember the numbers exactly. Hmmm. I could have sworn I had an SRPP circuit SPICE'd somewhere...

I'll grub around for it and see if I can offer you a better idea then.

[amptramp]

A µ-amp stage works best when the resistor between the upper source and the lower drain is equal to the reciprocal of the transconductance of the upper transistor.  This equalizes the upper and lower swing.  In the geofex example, the ideal transconductance would be 1/1000 = 1000 micromhos.  The thor example is best for 1/390 = 2564 micromhos.

Remember also that unlike a voltage amplifier stage, the µ-amp requires a load to operate properly.  The totem pole structure will force the upper current to be equal to the lower current if there is no load.  The difference between upper FET current and lower FET current must go through the load.

[sault]

So I SPICE'd out that whole second half of the Thor pedal, and I came up with what I think is a pretty decent idea of what's going on.

The SRPP is giving a gain of about 7x, or about 17 dB. The opamp is giving a gain of about 1.6x, so the two together are about 11x (20-ish dB). The opamp presents a high impedance, so no gain is lost in the SRPP.

That much gain in the Jfet will hit the rails (well, +/- 3ish volts) pretty easily. If you're looking for something a little subtler, ie in a different circuit, you might want to consider lowering the gain a little.

You can mess with the gain in the SRPP a few different ways. Lowering the bias resistor cuts gain - at 470k, for instance, you've cut its gain in half. Same goes with the load - reduce the load below about 200k and you cut your gain pretty quickly.

So anyways, if you want some tonal color, sure, match the resistor values closely and feed it with a high value Vref resistor, and keep in mind that you can always cut the gain if it gets to be too much.

I second the motion about the FDD, too, even though I don't know how to implement it... isn't it basically a resonant filter in a feedback loop? Something like that.

[dthurstan]

Hey cheers for the replies. I'd read that GEOFEX article b4, but it makes more sense now. Yeah I definitely want to lower the gain as I want a range of sounds from cruch to balls to the wall rawk !

I should get a decent load imp from the condor i/p. I was gonna leave out the jfet on the condor part, but I don't think I can now coz of the loading. I can't find anything on how to implement the FDD or patent details. A resonant filter in a fb loop is that a notch then?

Cheers for all the tips. I should post a schematic soon...

[sault]

Okay, so I've managed to find a schematic for the FDD. I want to see if I can SPICE it out correctly, so that'll be a project for today. From what people have said about it, the purpose is to raise the output impedance - this increases how it interacts with the speakers.

Let me briefly explain that... A high output impedance increases the roll-off you get from a cheap high-capacitance cable, right? If you've got a 250kΩ output impedance coming out of the guitar, and 20 feet of cable with 600 pF of capacitance,  250kΩ and 600pF = roll-off of 1khZ! Contrast that to the low impedance coming off of some active pickups, for instance... maybe 25kΩ? That plus 600pF = 10kΩ. Far, far less impact, especially considering that frequencies above 3khZ are generally attenuated anyways.  (a brief non-technical article on cable capacitance)  http://www.soundonsound.com/sos/nov09/articles/guitarcables.htm

So in a general sense, high output impedance = more of an interaction with filters and other EQ that follows. Speakers can be thought of as complex filters with impedance that varies by frequency... an 8Ω speaker means an *average* of an 8Ω response - at some frequencies it could be in the 10's or 100's of ohms, at others nearly zero!

If the FDD control raises output impedance, it increases the effect that the speaker has on tone. A low output impedance = flatter response, a high output impedance = frequency response that interacts with the speaker. The term "damping factor" is the official term for this, btw. (Wikipedia : http://en.wikipedia.org/wiki/Damping_factor , a page with a handy table showing the relationship : http://www.audioholics.com/education/amplifier-technology/damping-factor-effects-on-system-response/damping-factor-effects-on-system-response-page-2 ).

This isn't always a good thing, btw, it depends on the speaker. Sucky speakers, or speakers that aren't ideal for guitars, will benefit from a higher damping factor (that is, lower output impedance for a flatter response to minimize that non-ideal factor), while other speakers (like a Celestion G12T-75!) may benefit from a lower damping factor/higher output impedance.

Sorry, trying to keep it non-technical. I could talk about this stuff all day!

So to wrap this reply up in a little bow, the FDD appears to raise output impedance and adds a bit of EQ. I'll SPICE it and see if I can learn anything from it.

[sault]

I was gonna leave out the jfet on the condor part, but I don't think I can now coz of the loading.

The Jfet would be good to keep in, especially if you patch into to the cab sim. That way both the Thor circuit and whatever external circuit will both see a decent impedance. Secondly, though, you're going to see some insertion loss going through that EQ section... having some gain to offset that loss is a good idea.

I'll be interested to see your schematic.

[dthurstan]

That great. I actually did a module at uni called electro-transducer design. So I'm with you so far, however I think I'd need to dust off the old notes if this gets anymore technical .

Would a higher o/p impedance have a similar effect to the loading down of the mu-amp o/p? So if I remove the 1st fet stage in the condor there would be more interaction with the 1st filter stage?

http://www.runoffgroove.com/condor.html

This is really interesting, when I first got into effects building I spent a lot of time learning about impedance, buffering and the importance of high i/p and low o/p impedance. Now your telling me thats bad!  

PS Just seen your reply re jfet stage in the condor. I have a feeling this circuit is getting more complex than I 1st imagined.

[sault]

I'm sorry if I'm making this more complicated than it needs to be.



This is what I'm talking about - this is good design :

                           Thor  --------\
                                                 ---->  Jfet  ->  rest of Condor cab sim
Direct Input to Cab Sim -------/



The comments about high output impedance were only meant in regards to the FDD circuit wrapped around the power amp section of the Marshall, the section directly driving the speaker - nothing else! High output impedance anywhere else is usually a bad thing, especially when it's going into a filter section - you'll throw off the filter!

To restate - high output impedance can *sometimes* sound good in a power amp, but it depends on the speakers.


Sorry for going off on a tangent and not being clearer about what I was referring to. I have a bad habit of doing that when I get enthusiastic about something!

[dthurstan]

Your not making it more complicated, Your input is really helpful, cheers.
So is the FDD circuit really only useful when i/p into a real power amp, not something to be incorporated into an emulation of a power amp stage?

Here is a link to the schematic I've done so far, the amount of time I spent cutting and pasting I should of just redrawn it .



Sorry for the size it's a bit difficult to read.

So far it has the bassman preamp stages with 3 band eq, a presence control (from Mike McCardle's The plexizer) the SRPP stage and the Condor. I have to work on the presence control and where the SRPP & the Condor meet. Also I was going to put the volume between the SRPP & the Condor but I'm not sure. I might just stick it on the end of the Condor.

[amptramp]

You need to reduce the resistance of the 18 K series gate resistor and the 1 Meg gate resistor that goes to ground in the stage following the µ-amp stage.  The µ-amp is usually used as a voltage amplifier stage, but look at where the current goes:  if the input to the stage is high, the bottom transistor increases current and the upper stage reduces current and vice versa if the input goes low.  But this is a totem-pole stage - with no load, the upper and lower current must be the same because they are in series.  The output load must be low enough to absorb the difference in current between the increased and reduced current or the output will simply slam against the upper and lower rails.  The µ-amp gives you a nice clipping distortion, but it does not work as a linear stage until there is a low enough load impedance.

[sault]

... which is exactly what we've been talking about. Drop the load resistor, reduce the gain. As we mentioned earlier, that might need to happen - there is a pretty decent amount of gain in that stage, probably a little too much for just "color".

Okay, so I've been a little SPICE monkey lately, so let's see if I can illustrate why I've advocated pulling the bias resistor down instead of the load. So this is the schematic as is - 0.1 volts in, so something like 30 dB gain.... lots! Even a tenth of a volt hits the rails.



Here is what the output looks like with a lower load. Yes, I swapped the cap and resistor so I could get a 2nd order HP out of it. Gain has dropped to a more reasonable 22-ish dB. Notice that even with a lower load, you've still got significant distortion!



Okay, so here was my contribution - lowering the bias resistor to 220k. Again, gain is down to around 22-ish dB, but see how much cleaner the waveform is? Far less distortion, very linear... and this is going into the 1M load! So really, the "non-slamming into the rail" distortion is highly dependent on the value of that bias resistor.



Finally, I figured we could try something somewhere in between... Here I've dropped the bias resistor to 470k, and put both resistors in the 10's of k. Gain drops to around 20 dB, which can be pretty manageable. Kept the 2nd order highpass below 80hz, and struck a decent balance between linear and a bit of non-linear "color".



It's probably worth noting that these are all from a tenth of a volt. More voltage coming in does mean less linearity - these are Jfets, after all! I started to get noticeable non-linearity around a third of a volt or so. So the "color" is definitely going to depend on how hard you hit it...

[dthurstan]

Ok, it's been a while. Sault those spice calcs are great. It's a great way of explaining whats going on in the circuit. So I've tried to take that on board when I've made up this schematic;



I thought reducing the bias resistor gave the best results. It also shows how I want to connections to work. I used the jfet gain stage in the condor as the volume for the o/p and also for volume control when using as a stand alone cab sim.

[amptramp]

I notice you have C10 returned to the top of R20 rather than the bottom.  It may seem like a small point, but that changes the operation of the circuit entirely.  The idea of the µ-amp is thtat the drop across R20 provides the signal input for Q5, the upper transistor on the totem pole.  As shown, it is just a load that might be operating as a constant-current load.

[sault]

I'm noticing what might be an issue with the cab connection. As is, there is no coupling capacitor between the mu amp and the cab sim, and the 10k resistor comes after the cab insert. That changes the input resistance that the cab input will see. A better way to do this would be



The benefits : coupling capacitor for both the output of the Thor and the input of the cab sim insert. Leaving the 1M where it is keeps the impedance for both, but moving the 10k before the insert means that you keep the voltage divider/impedance that the Thor sees, but doesn't mess with the impedance that the cab insert would see.

Okay, next issue that I see is the 1k pot on Q6 with the "volume" label on it. This isn't a volume control - it changes Q6's bias, and would function more as a gain control than anything else. It is useful in the original circuit because the passive filter immediately after it has a pretty significant insertion loss - nearly 30db - so using some boost to compensate isn't a half-bad idea.

My personal suggestion would be to put a buffer in first - that way you can have a volume out and the cab sim in parallel, a la this super quick mockup... here, J2 is your Q6.

[dthurstan]

@ amptramp cheers I'll sort that out for my next post of the schematic.

Sault cheers for the help. I thought that 1k pot won't quite work but I was skimping on the components  .

I just realised that this isn't really about the Thor any more as the title would suggest.

I'll post an updated schematic soon.

Cheers

[sault]

It's all good. You've picked two different schematics here and are trying to make them work not just together, but with extra functionality.

I'm glad that I can contribute.

I didn't think about it last night, but coupling capacitors between the buffer that I proposed and both the output stage and the cab sim would probably be good. As far as exact values I'm not 100% sure, depends on the value of the volume control at least. Caffeine isn't working right now, can't think... Ummm, well, Jack Orman's article on buffers has his coupling capacitors at 10 uF. Probably don't need to be that high, though, probably a fraction of that going into the Condor.

http://www.muzique.com/lab/buffers.htm

[dthurstan]

I've finally got back to this project after working on something else for ages. I breadboarded the Condor opamp section over the weekend. It sounds great. I then started added stages to it to work out how to get the i/ps & o/ps to interact. Heres the circuit;



So you can plug into the cab i/p the signal is boosted and you have a volume control for the speaker simulator. You also have the direct out through the buffer and the volume for standard distortion box operation. The mu amp provides lots of gain to get over the passive filter at the start of the speaker sim. Im not sure if I'll need some a cap from the cab/boost o/p?

Cheers to everyone who helped me out and cheers and Sault for the spice sims they really helped. I just need to buy a load of pots and I should have this thing up and running.

Dave

[Rutger]

 I'm planning on doing something similar, and I have a question referring to Sault's Spice explanation.
What about the gateresistor of the first mu-amp stage? In different mu-amp designs you can find different values (the mini-booster uses 10M!). What happens when you change that value, the same thing as with the 2nd stage?

My plan is to connect a ROG Peppermill (http://www.runoffgroove.com/peppermill.html) to a mu-amp stage, for use as a preamp (the volumepot comes after the mu-amp). The mu-amp is there for the extra gain/volume, and the push-pull character of the stage is a nice plus. I don't want the mu-amp to clip too soon, for the Peppermill is there to give me most of the overdrive I need. My plan therefor is to run the mu-amp on 18V (9V ref.) while the Peppermill runs on 9V.

First I thought just to use 1M for the gateresistor, but got confused by all the different values. What value should I choose to keep the mu-amp clean, at least for the first 3/4 of a turn on the gainpot?