puzzling speaker-connection trick (?) in old Peavey gtr-amp

[clintrubber]

Just curious what the Peavey-designer had in mind when connecting the bottom-pin of R32 (2k7) to the positive speaker terminal ?

It doesn't function as some kind of current-feedback or such ...
(for modding to mixed mode feedback: insert 0.33 Ohms between the speaker and ground, then connect the in-between node to the bottom-pin of C18 (22uF). This Backstage-version doesn't have this; later models did.)

... the only thing I could think of is to disable the power-stage when the speaker is disconnected, but as an unloaded solid-state power-amp isn't a problem I don't really see the need for this. This is not a battery powered amp.

The R32 to speaker-connection results in passing of some 8mA DC through the speaker, no big problem, but why ?

Any other ideas ? Just curious.

Bye /thanks
 

 

[PRR]

It's a boot-strap.

You commonly find the Vas stage's collector resistor split in two and the junction driven from a unity-gain buffer (the output stage). This increases Vas load impedance; but more importantly, "kicks" the node past the supply rail.

If you know the load is a low DC resistance, and does not mind few-mA DC current, and especially if it already has a capacitor, you can use the load as the bootstrap source.

This is VERY common in $2 phono amplifiers and cheap radios of a certain age. Less common at the dozens-Watt level, but still valid.

 

[PRR]

You might instead wonder why R26 is used.

Also why C16 R24 set a ~~200Hz bass cut (how small was this amp?).

Also odd to find *four* output devices at this power level. 50W/pair was practical long before this. Musta been some oddity of pricing or inventory led to 12W/pair.

 

[clintrubber]

It's a boot-strap.

Thanks for pointing out, being more scattered over the schematic I hadn't realized it being a bootstrap (no excuse though  )

 

[clintrubber]

You might instead wonder why R26 is used.

Agreed, yet it remains stubbornly present in later incarnations of various Backstages.

Also why C16 R24 set a ~~200Hz bass cut (how small was this amp?).

It's indeed bass-robbing...  10" speaker, open combo.  Dunno, but something comes to mind: when for e-bass I switched from a Peavey to an SWR amp I didn't knew what happened... all that bass!  Might be company policy ?

Note that when I toyed around in simulator for converting this to mixed mode feedback, the effect of the ~100 Hz resonance peak (10") didn't 'develop' until indeed I increased C16 in value. In other words: it'd be a bit silly to want to have that bump AND still filtering so much bass.

Also odd to find *four* output devices at this power level. 50W/pair was practical long before this. Musta been some oddity of pricing or inventory led to 12W/pair.

This must be to justify the bragging in the Operating Guide about FOUR extremely rugged power devices coupled with FOUR hefty 5 Watt emitter resistors.  Later Backstage incarnations use two (Darlington) power devices though.

FWIW, the quiescent current is modest though (total of 7 mA through the power-devices) - but again, later versions use four stacked diodes (for 18mA through the power evices in total).

 

[JohnRoberts]

You might instead wonder why R26 is used.

Agreed, yet it remains stubbornly present in later incarnations of various Backstages.

Also why C16 R24 set a ~~200Hz bass cut (how small was this amp?).

It's indeed bass-robbing...  10" speaker, open combo.  Dunno, but something comes to mind: when for e-bass I switched from a Peavey to an SWR amp I didn't knew what happened... all that bass!  Might be company policy ?

Note that when I toyed around in simulator for converting this to mixed mode feedback, the effect of the ~100 Hz resonance peak (10") didn't 'develop' until indeed I increased C16 in value. In other words: it'd be a bit silly to want to have that bump AND still filtering so much bass.

Also odd to find *four* output devices at this power level. 50W/pair was practical long before this. Musta been some oddity of pricing or inventory led to 12W/pair.

This must be to justify the bragging in the Operating Guide about FOUR extremely rugged power devices coupled with FOUR hefty 5 Watt emitter resistors.  Later Backstage incarnations use two (Darlington) power devices though.

FWIW, the quiescent current is modest though (total of 7 mA through the power-devices) - but again, later versions use four stacked diodes (for 18mA through the power evices in total).

Life is too short to ponder the merits of old "value" (read cheap) peavey guitar amps.

Three diodes to bias up a four drop output stage seem starved for class A current. 4 diodes will be more appropriate but check for thermal stability. The 0.68 emitter resistor seem pretty large.  With only 40V total PS, not much power on a good day. 

I vaguely recall talking with Jack Sondermeyer back in the late '80s about a cheap modest power guitar amp, that used darlington output power devices.  Since it was against policy to use bias trims, these darlington output amps were notorious for running starved when cold (low bias current so audible crossover distortion).  Peavey dealers living in cold states, learned to let them warm up, before demoing them to potential customers. 

At the time I was messing around with an automatic output stage biasing scheme, so I modified one amp to show jack (I don't recall the model some small crappy amp.). My bias scheme worked,  but Jack yawned, and kept doing what he was doing.  You don't fix it if it ain't broke.

JR

 

[clintrubber]

Thanks John for the inside story, interesting to read & hear about the no-trim policy & the precooking.

Just some fun & trying some things here with these amps (tonight I'll play the 'real amp' again).
These 'first version'(?) amps with those Russian Space Age knobs look pretty funky though. Perhaps  nostalgia.

One or two Backstage-incarnations later there are indeed Darlingtons & also 4 diodes. For fun I'll add one to this Backstage 30, perhaps also glue the diodes to the heatsink & further check for stability. And consider a trim.

And also beef up the heatsink (there are currently just two modest thin L-brackets for cooling
- initially I even thought these were the same type of L-bracket as those loved/hated side trims on the speaker baffle ;-)  (if anyone still has some of these, please let me know  http://groupdiy.com/index.php?topic=63032 )

Out of interest, could the automatic biasing be one of your patents ?

Best regards

 

[JohnRoberts]

Thanks John for the inside story, interesting to read & hear about the no-trim policy & the precooking.

Just some fun & trying some things here with these amps (tonight I'll play the 'real amp' again).
These 'first version'(?) amps with those Russian Space Age knobs look pretty funky though. Perhaps  nostalgia.

One or two Backstage-incarnations later there are indeed Darlingtons & also 4 diodes. For fun I'll add one to this Backstage 30, perhaps also glue the diodes to the heatsink & further check for stability. And consider a trim.

Peavey would typically  drill a hole through an L bar extrusion and route the tracking diode loosely through the hole.  It doesn't need glue to track, but IIRC the darlington power devices were problematic regarding temperature compensation with simple schemes.

And also beef up the heatsink (there are currently just two modest thin L-brackets for cooling
- initially I even thought these were the same type of L-bracket as those loved/hated side trims on the speaker baffle ;-)  (if anyone still has some of these, please let me know  http://groupdiy.com/index.php?topic=63032 )

Out of interest, could the automatic biasing be one of your patents ?

Best regards

No, when I came up with my auto-bias scheme, my boss at the time wouldn't let me use it, out of fear that it might not work.. (that wuss  ). My plan was to first use it in a modest 35W amp used inside a Karaoke unit I was designing. I figured that would give it a good production test before rolling it out into other higher power and higher volume products.

While I prepared a confidential disclosure in case we wanted to patent it, I never used it in a product, so didn't pursue a patent.

I think I have discussed this circuit before... generally i hang a long tail pair across the positive and negative emitter degeneration resistors. At idle with reasonable resistor values there will be several mV across that string. I had a couple different approaches to build in the several mV voltage offset into the LTP so it would be at equilibrium for a modest steady class A current. (one approach was running the two devices in the LTP at different current causing a different Vbe, another is to add a small resistor in series with one emitter running same collector current.. IIRC I think I favored the second approach as being more stable thermally)

I worked like a charm (like I mentioned I tested it on one of the darlington guitar amps). My only concern was that it might introduce distortion , from fighting the drive circuitry between the forced class A bias point, and saturation when the signal drive over-rides the active bias, but this is all inside the NF loop, so looked good in initial testing. 

This was decades ago, and since then I suspect  some other variants on this have been developed by others. It seemed like a good idea to me.  8) but I like all my ideas.

JR

 

[PRR]

R26 can be justified as protecting Q7.

*When* the outputs go short, the drivers try to drive the speaker. Fail, and their drivers take the strain.

However the only protection on the power devices is over-provisioning. When they do go, Q8 is very likely to try to pull-up Amps, and fail. Now Q7 tries as hard as it can to wake-up Q8. R26 limits flow in Q7 to 2mA, peak heat 0.010 Watts, Q7 won't fail.

But it may be an overlooked habit. What's the difference 8 deaths or 9 deaths?

> 10" speaker, open combo.

Yes, 200Hz bass-cut makes sense. A Ten in a small open back will flap/slap, unless you make the spider absurdly stiff, or cut the bass. Peavey would have sold you a substantial BASS amp. With one of those in the band, the guitarist does not *want* to go below 200Hz, but keep all his energy up high above the bass.

And in the showroom this short-bass gives the buyer a reason to try the higher-price models with bigger cones and boxes (and higher price, so another penny profit).

 

[clintrubber]

I think I have discussed this circuit before... generally i hang a long tail pair across the positive and negative emitter degeneration resistors.
...
IIRC I think I favored the second approach as being more stable thermally)

Thanks for sharing. It'd be a DC-servo type loop I imagine, so wouldn't a sufficiently low BW prevent any nasties in the audio ? At least for something like a gtr-sgnal.

It seemed like a good idea to me.  8) but I like all my ideas.

JR

 

[clintrubber]

R26 can be justified as protecting Q7.

I feel this resistor-riddle has been itching you, I like & appreciate it. Sounds like a possible reason, but like you say, adding it won't make this design totally failsafe.

> 10" speaker, open combo.

Yes, 200Hz bass-cut makes sense. A Ten in a small open back will flap/slap, unless you make the spider absurdly stiff, or cut the bass. Peavey would have sold you a substantial BASS amp. With one of those in the band, the guitarist does not *want* to go below 200Hz, but keep all his energy up high above the bass.

That's how I thought to have understood Peavey-the-brand when I switched from a Peavey to an SWR bass-amp back then: put the Watts where they matter most, and don't cry too hard about the robbed bass.  Could be a justified approach for enough cases.

Just for fun I attached a quick drawing which could serve as a Bring-Back-The-BassTM when connecting an external cab i.s.o. the internal speaker. 
(the ground-connection details, like proper location, & when the mentioned mixed-mode feedback has been added are left as an exercise to the reader for now)
This hookup might perhaps not be too relevant for this amp, but now I am going to have a look  at where my VibroChamp (open, 8"spkr IIRC) starts to cut bass. 

Bye

 

[JohnRoberts]

I think I have discussed this circuit before... generally i hang a long tail pair across the positive and negative emitter degeneration resistors.
...
IIRC I think I favored the second approach as being more stable thermally)

Thanks for sharing. It'd be a DC-servo type loop I imagine, so wouldn't a sufficiently low BW prevent any nasties in the audio ? At least for something like a gtr-sgnal.

it tried to regulate the voltage drop across the emitter resistors to insure a minimum class A current, but didn't have enough power to stop it when the audio signal commanded more output current.

I didn't get to bench test it much, but it looked and sounded reasonably clean.

If I decided to take it higher power and higher fidelity i would have tweaked it further on the bench, but since it was a local feedback loop, inside the overall global negative feedback, it was good enough for those mid-fi applications. 

JR

 

[PRR]

drawing which could serve as a Bring-Back-The-BassTM

Cute.

I do wonder if plug tarnish might inject output back to input.

A simple "DEEP" switch throwing another 0.22uFd across the 0.047u is another method.

 

[clintrubber]

it was good enough for those mid-fi applications. 

Can imagine that, especially when each additional part needs to be justified.

Must say that many discrete solid state power stages without some more protection for keeping a speaker alive keep surprising me (sometimes there's a speaker-fuse, but rarely some overcurrent-protection). One would expect the speaker is the most important part to keep alive. Guess the companies know with what they can get away with.

Best regards

 

[clintrubber]

Cute.

I do wonder if plug tarnish might inject output back to input.

Indeed,  a certain risk there. After seeing the cathode-connections to J1 in the attached pic (Rivera), I figured one might get away by more fun&tricks with switching connectors than I previously assumed.  Anybody feeling comfortably with this ? I guess it can indeed work, more of their schematics show it.

A simple "DEEP" switch throwing another 0.22uFd across the 0.047u is another method.

That'd require yet another hole to be drilled in the chassis (or a pull-pot), but indeed is more flexible.

Best regards

 

[PRR]

> protection for keeping a speaker alive .... One would expect the speaker is the most important part to keep alive.

Guitar speakers are remarkably tough. The "normal" waveform may be totally-clipped for a large fraction of a night. Thermal input nearly twice the "clean" power. Suspensions are very stiff and sturdy, unlikely to be damaged by excursion.

Yes, guitar speakers do die, as we all know from rescuing the survivors. But it is amazing how rarely that happens, even with to-rail failures of amplifiers. Often the failure turns out to be the low-price model of a line, where sturdier speakers were available standard at a higher price. You pay for what you get.

 

[JohnRoberts]

it was good enough for those mid-fi applications. 

Can imagine that, especially when each additional part needs to be justified.

Must say that many discrete solid state power stages without some more protection for keeping a speaker alive keep surprising me (sometimes there's a speaker-fuse, but rarely some overcurrent-protection). One would expect the speaker is the most important part to keep alive. Guess the companies know with what they can get away with.

Best regards

If anything breaks the customer blames the manufacturer before themselves, especially with value (cheap) gear.

Peavey is actually quite good about balanced engineering for total system reliability. If not with the first pass, large scale production refines the thought process by highlighting any weak links in the chain.

I recall one painful situation where a competitor's hot new power amp, had a bad habit of introducing DC into the output from asymmetrical clipping/current limiting that would trash speakers.  The word on the street was that the new XYZ amp was blowing away the peavey speakers because it was so powerful (not)... Peavey bit the bullet and paid a lot of underserved warranty claims caused by a rogue competitor's amp, because the customers don't care about finger pointing. The good news is the XYZ amp engineers got their act together after a while and stopped blowing up peavey speakers en masse.

I actually killed a bunch of brain cells thinking about amp reliability and good design discipline trumps trick circuits. It is often cheaper to increase voltage or current of a component than add redundancy with more complexity and more points of failure.

JR

 

[ricardo]

I actually killed a bunch of brain cells thinking about amp reliability and good design discipline trumps trick circuits. It is often cheaper to increase voltage or current of a component than add redundancy with more complexity and more points of failure.

I'll second that.  If (heaven forbid  ) I get involved in PA design again, I'd use a certain 'fail safe' topology which would shut down by itself without extra circuits.

It won't give 1ppm THD at 20kHz but sensible behaviour on failure easily trumps that and sensible behaviour on overload certainly results in better performance in DBLTs  8)

 

[clintrubber]

Speaker is indeed still healthy, now let's see which component(s) keep(s) this powerstage from properly functioning. Debugging tonight.

Apart from that, anybody any feelings w.r.t. that Rivera powerstage disabling by means of the inserted speaker-plug ?  (pic attached)

Have a good weekend

 

[clintrubber]

A simple "DEEP" switch throwing another 0.22uFd across the 0.047u is another method.

Let's add that larger cap
... and add it here for completeness, as well as being the easiest location for myself to find it back...

Too lazy to drill holes or find push/pull pots,
replacing the TS-line-output jack by a TRS-insert + larger cap seems the simplest,
see drawing
(ignore T & R physical locations as drawn, location of Ring is obviously closest to Sleeve)


plug in TS-cable --> SEND, & PWR-Amp muted
plug in TRS-cable --> INSERT (or PWR-Amp input on Ring) with increased LF-BW
plug in (angled) TRS-plug with T+R shorted --> increased LF-BW

Click-suppression: let's add some MOhm-resistors as well to keep those floating capacitors a bit more defined