Fender Frontman 65R

[nickb]

* the circuit around Q21 and Q22 is just a thermal overload protection, I hope PTC RT1 is in direct contact with the heat sink.

Take a look at Q22, it connects between ground and -16V. Its emitter connects to GND. There is a cap between the base of Q21 and GND. So there is no DC connection between the PTC and the diff amp CS and so can't possibly work. That's drawing error no 1 IMHO.

Edit:
* D15/16 wonīt make U7a distort because they are neither in U7a feedback loop (should go from output to -in) or, say, from output to ground.
I also thought so at the beginning but now I see those are just the differential pair input to input protection .
Why on an input which will never ever see the outside World?

This is how I see it - you can tell me where I'm going wrong. D15 & D16 do connect between the output and -ve input of U7A, there just happens to be 3.9K resistor in series. There is a feedback divider formed between that and the input impedance of the diff amp (base of Q7) and the primary negative feedback network. The output of U7A has to get above +/- 0.65V to provide any local feedback so it's output is distorted. The output of U7a also connects direct to the input of the diff amp the base of Q6. The drive to the base of Q7 is essentially the same as the input and so has the same phase as the drive to Q6. It makes no sense to me.

In England I've heard the saying "about as much use as an inflatable dartboard".

[Enzo]

Useless as tits on a boar hog. (Or locally, on a bull)

[Enzo]

nick, I am also puzzled by that lack of DC connection. I looked at the same circuit in some other models, and it is drawn the same. SO they either cut and pasted an error, or we are missing something.

Look at the Rumble 100, the circuit is about the same, but it has one important thing, a wire from the collector of Q22, to the base of Q21.
http://bmamps.com/Schematics/fender/...Schematics.pdf

I suspect that is what it intended here too.

[J M Fahey]

Take a look at Q22, it connects between ground and -16V. Its emitter connects to GND. There is a cap between the base of Q21 and GND. So there is no DC connection between the PTC and the diff amp CS and so can't possibly work. That's drawing error no 1 IMHO.

Yes , just checked the FM212 schematic which is prectically the same, and this one is missing the line connecting Q22 collector and Q21 base.
This permits Q22 turning ON with PTC hot so it turns Q21 OFF which turns differential pair current OFF.

This is how I see it - you can tell me where I'm going wrong.

Oh, I can tell NOBODY heīs wrong .

D15 & D16 do connect between the output and -ve input of U7A, there just happens to be 3.9K resistor in series. There is a feedback divider formed between that and the input impedance of the diff amp (base of Q7) and the primary negative feedback network. The output of U7A has to get above +/- 0.65V to provide any local feedback so it's output is distorted. The output of U7a also connects direct to the input of the diff amp the base of Q6. The drive to the base of Q7 is essentially the same as the input and so has the same phase as the drive to Q6. It makes no sense to me.

To me neither.
I donīt think those diodes will ever clip, not based on that weird dual op amp input stage I donīt know the purpose of, simply in an indirect way: in any Op Amp (the power amp is just a larger version of), differential inputs in theory are at exactly the same voltage/potential and in practice justa few mV apart , few as in less than 5 mV, 2 orders of magnitude below clipping point.
No matter what the designer did or wanted, he canīt beat Physics Laws, so that input stage "must do something" to keep differential signalb etween those limits.
So canīt prove what it does, but can be quite confident on what it can not ... better than nothing

In any case, IF it were a killer power stage, miles above anything else, I would use long nights and gallons of coffee to fully explain it ... as is itīs not worth the effort.

Again: not bad, FM212 has excellent cleans and so so distortion, nothing away from the standard Fender SS.

In England I've heard the saying "about as much use as an inflatable dartboard".

[nickb]

I donīt think those diodes will ever clip, not based on that weird dual op amp input stage I donīt know the purpose of, simply in an indirect way: in any Op Amp (the power amp is just a larger version of), differential inputs in theory are at exactly the same voltage/potential and in practice just a few mV apart , few as in less than 5 mV, 2 orders of magnitude below clipping point.

I think I've ( finally ) got my head around this. Just as you say, I was overlooking the affect of the main power amplifier which restores normality via the feedback and so the diodes are indeed just input range limiting. By removing all the non essential stuff I simplified it to this:



U1 is U7A and U2 is the power amp. Beta is the gain of the feedback network. Ultimately, it behaves just like a traditional feedback arrangement with gain 1/Beta. It's still not clear to me why U7-b is arranged the way it is.

[Gingertube]

Jaun and etc.,
Thank You for your insights - they helped heaps in my understanding of the circuit.
It begins to make sense.
Q14 and Q16 are indeed the bias networks for the output Darlington Pairs.
They are X2 Vbe multipliers to provide the 1.4V bias the the output emitter followers (Set by R95, R97 etc.) and if the designers were clever enough (which I suspect they were) to put Q14 in close thermal connection to Q15/Q18 and similarly put Q16 close to Q17,Q19 then they will act with separate, dedicated temperature compensation.

Vbe multipliers are pretty much a pure voltage source (very low AC impedance) so what AC signal appears at the emitters also appears at the collectors. That means what AC signal appears at the collectors is the AC signal from the Q10,Q11 VAS Output.

Q10 and Q11 being a push pull Vas (Voltage Amplifying Stage) rather than Q10 being the Vas and Q11 an active load for Q10 as you suggested. C51 and C52 would suggest that this is correct, and that these caps are the amps dominant high frequency pole for amplifier stability.

Note that the AC impedance to 0V at Q15 and Q17 bases (top of the bias Vbe multipliers) has been bootstrapped via C53 and C54 to keep that "load" impedance high. These are the actual loads for the VAS, when you take into consideration that the Vbe multipliers are an AC short circuit.
Q12 and Q13 are output short circuit protection as you say, limiting output currents to about 3 Amps max. with some additional Over voltage drive limiting on the Vas output, softened by R84.

The bits I still have trouble with:

I could not make sense of Q11/Q22 until that comment above about a connection missing from Q22 collector to Q21 base. With that it make sense, and note R113 and C61 then give a 2.2 second time constant for a startup delay. C61 has to charge up to D39 (10V) zener + Q21 Vbe voltage to get -10.7V on Q21 base for it to turn on, that then allows the diffamp current source to turn on. So explained.

That leaves the "weird" U7-B. And I still don't understand it.
It is establishing the DC and AC references for the diffamp -ve input (Q7) base. My circuit analysis skills fall short here .

The resistance/impedance to 0V,looking into the opamp output is about 10 Ohms. The opamp itself has a gain of 4 for frequencies up to when C46 starts to come into effect which is when its impedance drops to that R69+R68 = 320K. Which is 50Hz. Above that its gain is drops, eventually to 1.

What Am I looking for,
I want to go at this backward. If I know the output level (TP21 says 2.10VAC) for +ve diffamp input,Q6 base (TP19 = 37mVAC)., Then I need to find a gain of 56.
Actually a smidge higher since the output stage has a gain of just less than 1 (say 0.95) so I'm looking for a gain of say 60.
ASSUMING (a dangerous thing) that the -ve of the diffamp is only receiving feedback signals and no inverse drive then R70=680K means I am looking for about 11K impedance to 0V at the R65,R70,R71,R73 junction and I'm not seeing it??
I can see 27K via R71, C48,C49 but we would need another 27K (actually about 20K) in parallel. Maybe that is provide by U7-B.
Problem here is I also see R73 = (1K8) back to 0v via the negative speaker output (0.1 Ohms above 0V). This 1K8 will be bootstrapped a little due to voltage developed across R107 (the negative current feedback) but not nearly enough to get a value of 20K. (more like 2K).

The fact that I can't make sense of this suggest to me that there is a fundamental problem with my analysis ,the most likely being that one of those critical nodes is a summing junction (virtual earth) and therefore I should be looking at feedback currents rather than voltages.

Possible Light Globe Moment: The whole Amp is differential amplifier, just like an op amp, so its -ve input, the base of Q7 is a summing junction and a hence a virtual earth.
Old Farts guidance to newbies: A virtual earth is a point of AC=0 volts and all of the currents flowing into that node MUST sum to 0. To analyse the circuit you have to think in terms of current not voltage. That is, in terms of the voltage and the series resistance to that node.

If we adjust our view to accommodate this current centric view then the U7-B starts to make more sense. Below 50Hz it actually applies -ve feedback to cut the effective input signal, above 50Hz when its gain drops to 1 then it simply cancels the input signal.

Any insights as to this? (and thanks for the learning experience).

Cheers,
Ian

[Dave H]

That leaves the "weird" U7-B. And I still don't understand it.

The resistance/impedance to 0V,looking into the opamp output is about 10 Ohms. The opamp itself has a gain of 4 for frequencies up to when C46 starts to come into effect which is when its impedance drops to that R69+R68 = 320K. Which is 50Hz. Above that its gain is drops, eventually to 1.

It looks like it will operate as a unity gain buffer to me. R68 is the feedback resistor. R69 is inside the feedback loop.

[J M Fahey]

Jaun and etc.,
Thank You for your insights - they helped heaps in my understanding of the circuit.
It begins to make sense.
Q14 and Q16 are indeed the bias networks for the output Darlington Pairs.
They are X2 Vbe multipliers to provide the 1.4V bias the the output emitter followers (Set by R95, R97 etc.) and if the designers were clever enough (which I suspect they were) to put Q14 in close thermal connection to Q15/Q18 and similarly put Q16 close to Q17,Q19 then they will act with separate, dedicated temperature compensation.

Agree.

Vbe multipliers are pretty much a pure voltage source (very low AC impedance) so what AC signal appears at the emitters also appears at the collectors. That means what AC signal appears at the collectors is the AC signal from the Q10,Q11 VAS Output.

Yes.

Q10 and Q11 being a push pull Vas (Voltage Amplifying Stage) rather than Q10 being the Vas and Q11 an active load for Q10 as you suggested. C51 and C52 would suggest that this is correct, and that these caps are the amps dominant high frequency pole for amplifier stability.

You are right.
In fact I noticed that, I even wrote that it was similar to the classic Toshiba Mosfet Amp mirrored gain stage, bottom transistor driven out of phase, but to keep an already complex explanation more readable I edited that and just called it "active load" which it is, should have called it a "dynamic active load" , to differentiate it from "constant current load" which is the other active load I know.
The main idea was to separate it from the *two* conventional bootstrapped loads which are also present :
* R91 + R92 bootstrapped by C53 and
* R89 + R90 bootstrapped by C54
so in the long run designer did NOt improve from lowly bootstrapped loads to fancier active/push pull ones and in fact now needs TWO of them .
Just a personal rant: looks like designer wanted to create something unusual or revolutionary but at every step he has to go back to the tried and true solutions, he pulls something away and then has to add it back in a more complex (should I say clumsier? ) way.
Reminds me a lot of designs found at DIY Audio, where many "just want to do things a different way" ...... "just because they can" .....
Not surprised at what I find there, but in general MI designers have been practical, feet on the ground types .... maybe because of the tradition started by Fender, Ampeg, Marshall, Vox, Laney, Peavey, Crate, Acoustic, etc.
Even the advanced Eden amp we have in another thread, does all the magic in the premp (and speakers) , but the power amp is basically a HUGE Op Amp which linearly amplifies what is fed to it.
This amp (I guess) tries to add some tricks inside, maybe to reach a "tubey" sound? , or maybe trying to reach VERY low distortion, not what guitar players exactly demand.

There IS a DIY Audio project where designer added an op amp in front of a regular power amp and applies NFB along the whole enchilada, the claimed improvement being that instead (just quoting from memory) of 0.0054% distortion or some similar low value, now he has 0.000054% because Op Amp has 100X gain so now NFB is 100 times higher (because total gain is kept the same).
My gripe with such numbers is that they never ever measured such low distortion (which is impossible to do at home and very difficult in a very well equipped Lab) but simply numbers crunched by some simulation software.

I see some of those trends here, including the : "look Ma!!!! no biasing network in the VAS !!!!!" which of course has to be added later, twice, and ... non adjustable.
- "look Ma!!!!! , push pull active load!!!!!" .... followed by TWO bootstrapped loads required to pass some current through the two bias transistors ... and so on.

I do not simulate, old dog learns no new tricks, I am from the:

era , can some of the younger members (cough cough) simulate what U7 does?
Just U7a/b alone , injecting signal through C42, sweeping it from 20Hz to , say, 10kHz, plotting whatīs seen at U7a output (pin 1) and at junction/node C47/R68/R69 .
Unloaded, I consider Q6 and Q7 base high impedance loads which wonīt affect results.
Oh, also disconnected from R70 and anything which brings NFB from the main amp.
I want to know what "flavour" does it add to NFB.
Might be something as pedestrian as limiting subsonics or, on the contrary, adding some bass boost for a fuller sound, maybe a bump at 90/100 Hz?.
Fender is known for their famous active filter at preamp input, not bad IF it were buffered but inadequate to be driven by a passive guitar pickup and circuit , I bet that it simulates or measures well ... if driven by a Lab generator that is.

Note that the AC impedance to 0V at Q15 and Q17 bases (top of the bias Vbe multipliers) has been bootstrapped via C53 and C54 to keep that "load" impedance high. These are the actual loads for the VAS, when you take into consideration that the Vbe multipliers are an AC short circuit.

Very true.

Q12 and Q13 are output short circuit protection as you say, limiting output currents to about 3 Amps max. with some additional Over voltage drive limiting on the Vas output, softened by R84.

Yes, R85 pre biases Q12 by passing current through R86 in parallel with R96 so its base is more sensitive, it turns on earlier IF output is near ground (short) so it limits current to low, safer value; now if output is near top rail, R85 gets out of the way and current limiting comes from R103 drop attenuated by the R96 R86 divider triggering Q12, so some 7 Amperes peak, same is repeated symmetrically on the lower side.
Very tight spec, a 4 ohm load there requires 8A peak at least so I bet it regularly clips when playing loud with real world speakers, maybe that is a wanted effect to make sound more "tubey" , I do something similar but in a WAY simpler way .

The bits I still have trouble with:

I could not make sense of Q11/Q22 until that comment above about a connection missing from Q22 collector to Q21 base. With that it make sense, and note R113 and C61 then give a 2.2 second time constant for a startup delay. C61 has to charge up to D39 (10V) zener + Q21 Vbe voltage to get -10.7V on Q21 base for it to turn on, that then allows the diffamp current source to turn on. So explained.

So you donīt have such trouble any more [smile] <-- Sorry, I already run above the Forum allowed smiley quota, so read this as such.

That leaves the "weird" U7-B. And I still don't understand it.
It is establishing the DC and AC references for the diffamp -ve input (Q7) base. My circuit analysis skills fall short here .

The resistance/impedance to 0V,looking into the opamp output is about 10 Ohms. The opamp itself has a gain of 4 for frequencies up to when C46 starts to come into effect which is when its impedance drops to that R69+R68 = 320K. Which is 50Hz. Above that its gain is drops, eventually to 1.

What Am I looking for,
I want to go at this backward. If I know the output level (TP21 says 2.10VAC) for +ve diffamp input,Q6 base (TP19 = 37mVAC)., Then I need to find a gain of 56.
Actually a smidge higher since the output stage has a gain of just less than 1 (say 0.95) so I'm looking for a gain of say 60.

As I see it, the basic NFB *voltage* is Vout * 680k/1k8 (voltage feedback) in series with Vout * 0.1 ohm/4 ohms (current feedback), so actual gain is Vout/Vnfb .
U7b modifies that but Iīm not sure about what it does, except that it works at low frequencies, since it has that .01uF cap from OUT to -IN

ASSUMING (a dangerous thing) that the -ve of the diffamp is only receiving feedback signals and no inverse drive then R70=680K means I am looking for about 11K impedance to 0V at the R65,R70,R71,R73 junction and I'm not seeing it??
I can see 27K via R71, C48,C49 but we would need another 27K (actually about 20K) in parallel. Maybe that is provide by U7-B.
Problem here is I also see R73 = (1K8) back to 0v via the negative speaker output (0.1 Ohms above 0V). This 1K8 will be bootstrapped a little due to voltage developed across R107 (the negative current feedback) but not nearly enough to get a value of 20K. (more like 2K).

See above.

The fact that I can't make sense of this suggest to me that there is a fundamental problem with my analysis ,the most likely being that one of those critical nodes is a summing junction (virtual earth) and therefore I should be looking at feedback currents rather than voltages.

I think thatīs the way, I also look at currents and sometimes when I mention voltages (such as: Vout * 0.1/4) I am actually saying (same but in other words): "voltage dropped across a resistor by a *current* supplied through a much higher value resistor"

Possible Light Globe Moment: The whole Amp is differential amplifier, just like an op amp, so its -ve input, the base of Q7 is a summing junction and a hence a virtual earth.
Old Farts guidance to newbies: A virtual earth is a point of AC=0 volts and all of the currents flowing into that node MUST sum to 0. To analyse the circuit you have to think in terms of current not voltage. That is, in terms of the voltage and the series resistance to that node

.
True.
I am often dismayed by the recurring post about: "such and such Op Amp must be bad, I scope its input and signal disappears, yet it shows strong, clean, amplified signal at its output, WTF?????" where they are trying to scope a virtual earth.
A similar one: in an Op Amp input power amp, "scope shows a very distorted signal at the output of the Op Amp, yet speaker output looks reasonably clean, WTF???" while in fact there you have the (ugly) *error* signal, the bits and pieces the output signal is missing to be a true sinewave and which, of course, are *anything* but sinusoidal.

If we adjust our view to accommodate this current centric view then the U7-B starts to make more sense. Below 50Hz it actually applies -ve feedback to cut the effective input signal, above 50Hz when its gain drops to 1 then it simply cancels the input signal.

If possible, please simulate just that gain block, as I asked above. Thanks.

Any insights as to this? (and thanks for the learning experience).

Personally Iīm here to learn , got invaluable info and insights here from lots of generous people, plus a Ton of hands on practical experience by you-know-who [wink] plus to me this amounts to going to the Gym every day (the mental one that is, the physical one is sadly neglected)

[Dave H]

can some of the younger members (cough cough) simulate what U7 does?
Just U7a/b alone , injecting signal through C42, sweeping it from 20Hz to , say, 10kHz, plotting whatīs seen at U7a output (pin 1) and at junction/node C47/R68/R69 .
Unloaded, I consider Q6 and Q7 base high impedance loads which wonīt affect results.
Oh, also disconnected from R70 and anything which brings NFB from the main amp.
I want to know what "flavour" does it add to NFB.

I don't think you can simulate just U7a/b alone disconnected from the NFB of the main amp because the main amp is inside the FB loop of U7a. With the NFB disconnected U7a will have no DC/LF NFB except for D15/D16 which is a 0.6V discontinuity. Ill try to simulate U7b alone. Its signal in is its non-inverting input and there's no resistor to ground from its inverting input so I think it's a non-inverting buffer.

Edit: Added simulation

Flat except for tiny dip of 1.8mdB @ 100Hz

[nickb]

I don't think you can simulate just U7a/b alone disconnected from the NFB of the main amp because the main amp is inside the FB loop of U7a. With the NFB disconnected U7a will have no DC/LF NFB except for D15/D16 which is a 0.6V discontinuity. Ill try to simulate U7b alone. Its signal in is its non-inverting input and there's no resistor to ground from its inverting input so I think it's a non-inverting buffer.

Edit: Added simulation

Flat except for tiny dip of 1.8mdB @ 100Hz

[ATTACH=CONFIG]41551[/ATTACH]

[ATTACH=CONFIG]41552[/ATTACH]

I think you need to add the load impedance of the diff amp to get realistic numbers. But really that is a detail. I think we know what it does, the real issue is why? A unity gain buffer would have done about the same with less components. What special characteristic was designer seeking? The buffer has about a 50K output impedance at DC that can only serve to add to the input offset.

[J M Fahey]

Thanks

You are right, there is no DC path between U7A OUT and -IN so it canīt work alone.

So I guess the full amp needs to be simulated

Wonīt happen in any nearby date

*Maybe* , but thatīs just a big unknown, with full gain and overall feedback that definite but tiny dip at 100 Hz gets larger ; the speaker , fed constant current, will have a large peak at resonance; this circuit might compensate that , leaving high frequency boost intact, but Iīm just thinking aloud here, Fender are not fools so there must be some reason for that, itīs just we didnīt go deep enough.

In any case, IF it were a KILLER amp, miles ahead of others, then it might pay to get deeper, but as-is, being a competent but average amp, no justification.

Thanks again

[Dave H]

the real issue is why? A unity gain buffer would have done about the same with less components. What special characteristic was designer seeking?

The whole power amp could do about the same with half the components They obviously hadn't heard of the KISS principle. How could you start with a blank sheet and end up with a circuit that complicated? Design by committee? How does the component count compare with a Fahey amp Juan?

[Enzo]

They didn't start with a blank sheet and come up with this. They started with a conventional amp, and tweaked and tweaked and tweaked. They were trying to get the loudest amp they could with the smallest power supply they could.

[Dave H]

They didn't start with a blank sheet and come up with this. They started with a conventional amp, and tweaked and tweaked and tweaked. They were trying to get the loudest amp they could with the smallest power supply they could.

How could all their tweaks make it louder? Once it hits the rails that's as loud as it's going to get.

[Steve A.]

They didn't start with a blank sheet and come up with this. They started with a conventional amp, and tweaked and tweaked and tweaked.

As you very well know (I think you are the one who told me this) the current line of Frontman amps were largely based on Fender SS amps going back to the early 90's and possibly even earlier, namely the SS amps designed by Rivera as I recall.

I really like the various Champ, Princeton and Deluxe SS amps*** built through the early 2000's before they added all of the digital FX. The earlier Frontman amps built in Mexico were pretty nice, too, with decent reverb springs.

At least for the smaller models like the FM25R the reverb springs in the Asian Frontman amps are cheap plastic crap to be avoided.

FWIW I love the earlier amps that used power amp modules so I don't have to troubleshoot those darned transistors...

Steve Ahola

*** Like the earlier 65 watt Princeton 112 and 112+ amps and the later Princeton 65, the preamp of which apparently evolved into that in the FM65 after borrowing some switching tricks from the Deluxe 90 (which is typically selling used for about $150 if you can buy locally to avoid s/h charges.) Don't expect the OD channel to compare with current generation analog SS amps like from Orange but the amps take pedals very well and are fairly easy to mod. If you are looking for a lot of power in a fairly light amp the 65 watt Princetons and 90 watt Deluxes are hard to beat (the 30 watt Champs are pretty good for smaller gigs.)