Help me understand this power amp design - please?

[LJ King]

Hi,

I am having a heck of a time trying to confirm (at least in my mind) if this power amp schematic works or not. I want to perfboard build a clone of the entire amp, but I would hate to find out that the schematic is wrong. I've done lots of research over the last two weeks and this is driving me nuts.

Here is the PA schematic section:



The amp is a "Mike Matthews Freedom Amp" produced by Electro-Harmonix during the early-mid 1970s. The single ended B+ supply of 60 volts was either AC derived through a minimal power supply (transformer, bridge rectifier, one electrolytic cap) and/or provided by 40 "D" cell batteries. Advertised as "100 watts peak" its RMS output is somewhere between 15 and 25 watts.

On the schematic, transistors Q5 through Q10 are indicated as either FS37001 or FS37003. I haven't been able to find a datasheet or crossreference for these numbers. However, there is an asterisk by Q9 and Q10 and a note that Q9 is replaced with a 2N6123 for the amps using more than one speaker. Since the 2N6123 is in a TO220 case and rated for 50 watts (with a 2N6126 complement) it makes some sense since they may have to provide more current to drive less than 8 ohms of the single speaker versions. So I'm assuming right now that I can find a suitable substitute using the 2N6123 characteristics.

Let me tell what I think about this circuit before I get to the problem as a mistake in my understanding may have something to do with how I got myself confused if this amp possibly works.

Q11 and Q12 is a complementary Class B output stage. No emitter resistors, but that shouldn't affect it from working. Q9 and Q10 is a complementary Class B driver stage.

Q7 and Q8 had me confused until I realized the only way it could possibly work was push-pull Class A. I assume it provides voltage amplification for the Class B drivers.

Q5 and Q6. Well obviously an input amp. Receives feedback from the output across the emitter resistors through R22 and R24. I think R27 and R28 provide a "centerline" voltage reference (~30 volts) at the junction of the emitter resistors (R21, R23). Q5 and Q6 also have to work as a phase inverter for the push-pull drive to Q7 and Q8.

Up to this point I am somewhat satisfied.

What is causing my problem is Q5 and Q6 - how are they biased? The diodes between the bases don't confuse me, I've seen those too many times like Q7 and Q8. I understand that this will set the bias between the two transistors at three diode voltage drops. I can even convince myself that the need for that third diode is to make up for the voltage drop across the series connected emitter resistors.

What totally confuses me is the lack of a DC path to establish the bias. The input is capacitively coupled. How can the base of Q5 be more positive than it's emitter when there is no source of a positive voltage at its base? How can the base of Q6 be more negative (i.e. less positive) than its emitter for the same reason? I'd be happy if there were a resistor from ground to the base of Q6 and another resistor from the base of Q5 to B+. But there isn't.

One thing that seems odd about the schematic is that the value of Q5s collector resistor isn't the same as the collector resistor of Q6. Everything else is so perfectly symmetrical with this circuit that this seems to stand out like a sore thumb, but I don't know what to make of it other than to assume Q5 needs more voltage gain than Q6 for some reason.

So, either this circuit won't work or I am totally not seeing something. I really hope it's the latter.

Any help?

Thanks
LJ

[teemuk]

Q5 - Q7 and Q6 - Q8 pairs are current feedback stages. The first transistor is the input, second the voltage amplifier. Since there are two of them the amp is symmetric. The asymmetric version of this input-VAS combination was pretty common before long-tailed pair input became as common choice it is now.

The bases of input transistors shoud be biased to a proper DC potential - I don't see it happening. Feedback scheme looks wacky... hmm....

Just to prove my suspicions about this circuit that looks a bit messed up right I draw the damn thing on spice and ran a sim....

An surprise: It doesn't work! Well, it does if you replace the input transistors with FETs that do not need a DC bias but guess what: The circuit also has less than unity gain. I suppose someone did a hack job drawing this thing out. Where did you get the schematic?

It sort of looks like the joining node of R21 and R23 should have a common reference - which, I suppose, would lead the complementary pair Q9 - Q10 to exhibit some gain as well (a pretty nice invitation for instability by the way). Well, I run another sim: Ain't happening either because bias is messed up. I guess I could tweak it to work but why bother since it's practically a redesign of the whole thing. I'd say forget this circuit.

[LJ King]

Well, that's just too bad. Good thing I was suspicious before I built it and wasted the parts.

I guess I'll check around and see if I can find a different schematic. I got this one from musicparts.com and it's supposed to be factory. It does seem to have an incomplete or maybe sloppy feel to it - the component numbers jump around, there are holes in the sequence, some components don't have numbers just values.

Thanks for the confirmation that it won't work as drawn!!

[LJ King]

Checked with other schematic places that say they have a Freedom Amp schematic, and apparently this one, dated 6/18/75, is the only version in existance. I don't understand that - during the early 1970s Electro-Harmonix would sell you the amp in kit form, so I would think there would be at least some dated before 1975.

I thought about getting a schematic for the "Dirt Road Special" as it should be basically the same amp. While I was searching I happened upon a almost too small to be useful DRS schematic posted on a German forum. After registering (necessary to view the schematic), I was surprised at how the power amp was almost identical!! Almost, but not quite.

Since it was too small I had to blow it up multiple ways to get the circuit values. Then I modified the original schematic with the differences:



From Q7 through Q12, the circuit is identical. All of the changes center around Q5 & Q6.

o The addition of R32 & R33 to the base of Q5/Q6.

o The preamplifier input is routed through a separate cap to the base of both Q5 and Q6.

o Q5/Q6 collector resistors are now identical values of 47K.

o The addition of C21 at the junction of R21 & R23.

I have much more confidence in this circuit actually working. Just two questions:

1) Does the addition of C21 provide the common reference you though should be at the junction of R21 & R23? I think my assumption is a voltage reference.

2) It seems to me that Q9 & Q10 are expected to provide some gain. I guess I don't understand why that would lead to instability.

Thanks!!

[teemuk]

1) Yes. With the addition of this capacitor the amplifier actually begins to have some gain and that base bias is also essential. I'm quite sure the original circuit was meant to be something like this.
2) I believe you find the most clearly stated answer to your question from here: http://sound.westhost.com/amp_design.htm#nogood

At least SPICE shows that the attached circuit is working: bias looks fine (no excessive power wasted anywhere) etc. Might have some problems in reality that yo'd have to compensate, i.e. the HF feedback is pretty much lacking. My only comment aside that and the issue #2 is the gain of 10. This seems pretty low. You will definitely need a preamp to compensate that issue.

[LJ King]

I was getting together a parts list to build this amp when I noticed something that got me to thinking.

The collectors of the output transistors are tied together.

The thought was - is it really necessary to ground the heat sink? If the heat sink is ungrounded, then the output transistors could be mounted directly to it without a mica insulator.

So... if I mount the transistors directly to the ungrounded heat sink, would it still be advisable to use a heat sink compound between the transistors and the heat sink?

[teemuk]

You don't need to ground the heatsink if you're sure it cannot be shorted. In fact, some copper mounting clamps that attach transistors to the heatsink may serve as a wondeful supply rail. Getting rid of the mica washer (which is both electric and thermal insulator) is also beneficial. However, in my experience insulating the heatsink from the rest of the chassis and making sure it cannot be shorted by accident is a lot more difficult than just putting the mica or silipad washer between transistors and the sink.

Like I said, getting rid of the mica insulator is nice (it's one considerable thermal resistance source less) but definitely use the thermal compound: It is essential for decreasing thermal resistance of the interface between sink and the transistor. Without compound you leave microscopic air pockets in between. Air is a very good insulator of heat, which is very bad. Just make sure the compound is not electrically insulative.

[LJ King]

Since I have this Dirt Road Special apart, I thought someone might like to see how it's put together. I have verified the second schematic posted in this thread and corrected a few capacitor values. If the input caps are showing something other than a value of "1", refresh your browser.



This is the insides. The speaker has been removed to provide some extra room to work with. The case and back panel are 1/2 inch particle board with the sides finger jointed in the corners. The front baffle is plywood. The wood strips used for mounting the back panel are stapled to the sides of the back. Similar strips are used to build up the front edge to hold the grille cloth and for the baffle to screw into. I'm sure that the small size helps, but it is a very sturdy case.

The power supply is in the upper left of the picture - transformer, open style fuse holder, bridge rectifier, electrolytic cap, and a .1μF (wired across the transformer secondary leads prior to the rectifier) are all wired together point to point with no brackets or spare lugs. The fuse holder is secured to the side of the cab, the transformer and electrolytic secured to the bottom with nuts and bolts through the cab. Everything else is "loose" with a few wires bundled with a plastic tie.

The "rats nest" of wires tie the power supply to the circuit boards. Three wires are separate grounds all coming off the main amplifier board. One wire supplies the B+, and there are two for the primary AC - one is routed to the power on/off switch and the other side of the switch comes back down to the fuse holder. There is no "chassis grounding" cap on the primary AC side, just the 1 amp fuse and switch.

All of these wires are single strand and a really thin gauge. They will break with the slightest nick, and several disconnected themselves as I was replacing the IC with a socket. I will be replacing all of them, and routing the AC lines separate from the others.

Of the two circuit boards, the smaller one at the bottom is the Small Stone phaser effect. The switches below mounted to the control panel are the on/off and Color for the phaser.

The preamp and power amp are combined onto the larger board. The single 4558 chip is for the preamp. The six TO-94 transistors and the two TO-3s make up the power amp. The two larger electrolytics are C20 and C21.

Both boards are secured to the control plate with the board mounted pots. There are also holes in the boards (left side in the picture) where screws go through the case, through a round, roughly 1 inch long plastic spacer, through the board and a nut secures the screw at the bottom of the board.



The output transistors and heat sink. Yes, the transistors are original and still working fine and I won't be replacing them. Notice the 7914 and 7926 dates.

They are mounted to the heat sink with screws on the bottom side of the board and secured with nuts. As discussed above, they are not insulated from the heat sink with mica, but do use heat sink compound. The heat sink does carry the "mid-point" 30 volts and of course is not grounded. The purple wire is one of the speaker connections.

I calculate the radiating area of the heat sink to be roughly 12 1/2 square inches. Seems a bit on the small side, even for 25 watts. It does get warm to the touch during operation. When mounted in the case the heat sink is quite close to the speaker, so one possibility is that speaker movement helps in cooling. With the exception of a 2 1/2 inch square hole cut in the back panel and covered on the inside with grille cloth, the amp is pretty much sealed when fully assembled.



This is the trace area of the board underneath the heat sink. No sockets. The emitters and base are connected with wire wrapped around the pins and the other end laying on a trace, both ends soldered of course. There are quite a few other connections made with wire soldered to the top of traces.

The trace under the screw to the upper left connects to the output capacitor (C20). This cap, as well as C21 are 35 volt caps. I replaced them with 75 volt units just because it was easy to do since I had the board out anyway (and it is a real pain to get the board out).

While there are definitely some improvements that could be made in the construction quality of this amp, it has proven itself over time to be very reliable and good sounding. Chalk it all up to very good luck, but it's not just "my opinion".

http://reviews.harmony-central.com/reviews/Guitar+Amp/product/Electro-Harmonix/Mike+Matthews+Dirt+Road+Special+112+Combo/10/1

As for the IC, I decided I liked the TL072 better than the 4558. Not any real difference in tone, or distortion or anything like that.The 4558 wasn't really hissssssy, but the TL072 was noticeably quieter.

[ryansteele]

Hello,

Please forgive me resurrecting such an ancient thread. However, information on this amplifier is criminally rare, and this thread has some of the best discussion to be found. I felt it appropriate to add further information regarding the transistors used in the EH Mike Matthews Freedom Amplifier here for this reason.

This information is based on my servicing/comparing two of these amplifiers. One of these is built using transistors with the obscure EH numbering in one section, and normal labelled parts in the other. The second amp is reversed in this regard. As such, I was able to ascertain what is what for those looking to service this amp, or try their hand at building the circuit properly.

[Schematic Label]	[Schematic Part]	[Stamped (Amp 1)]	[Stamped (Amp 2)]
Q1, Q3 NPN	2N5133	FS36999	2N5133
Q2, Q4 PNP	2N5138	FS37000	2N5138
Q5, Q8, Q9 NPN	FS37001	FS37001	2N3568
Q6, Q7, Q10 PNP	FS37003	FS37003	2N4354
Q11 PNP	2N4908	2SA627	S37005
Q12 NPN	2N3055	2SC1079	S37004

Firstly, this seems to confirm the longstanding suspicion that the "FS36999" (as also used in the early Big Muff) is in fact a 2N5133 NPN.

"FS37000" - 2N5138 PNP

On to the "FS37001" and "FS37003":

"FS37001" -  2N3568 NPN.
"FS37003" - 2N4354 PNP.

"FS37004" - 2N4908/2SA627 PNP.
"FS37005" - 2N3055/2SC1079 NPN.

Hope this information is of use to someone.

[Roly]

Welcome!

And thanks.   :dbtu: