Ampeg SVT3 Pro Bass Amp

[PRR]

> the "Tube Gain" control,... this thing has a strange delay built in

At the glance, C5 47u against R9 56k is a **2.6 second** time-constant.

Replace C5 with a 1u. It will come up/down quick. I think you will find why they didn't do that. The sudden change couples-through the audio path to make big thumps as you turn the knob.

Again some over-design. If they use a White Cathode Follower (V2), then V1 pins 678 is pointless. The WCF has the same "infinite" input impedance as the cathode follower.

And if there is ANY sand in the amp, then a WCF is excessive. If properly proportioned (you cut-off top and bottom resistor values so I can't tell) a WCF is about as dead-clean as any tube can be, no "color". A couple 80V transistors could do the job of three half-bottles, same sound, less heat. (Ah, but we are powering 12V heaters from a 65V rail. We need the extra heaters to waste the excess voltage. Where is the 36AX7 when you need one?)
_________________________

> no phase inverter

Complementary Symmetry (even if "quasi-") does not use a phase splitter. Both sides work as Followers. One drive point in the middle moves up and down.

(Since we need Bias, this "one drive point" is split by a bias-voltage device/scheme.)

This does require at least one P-type device. In modern production, we can afford half the devices to be P-type, so we usually find true complementary.

On BJT or FET, watch the direction of the little arrow. (And do not trust all designers and layout programs to get it right.)
_________________________

> Ampeg used to have a good name

That Ampeg is LONG dead.

"Ampeg" is now one of the Brand-Names for the St Louis company that Borged most of the old-line US brands. I suspect much of their design work is done by guys born after the Disco era, long after Real Rock (and Jazz) golden years. And even if they find a clue, decisions have to be about Product Features VERSUS Cost-To-Make, because Banjo Center takes 40+% of the retail price.

 

[CJ]

ok we stuck some new resistors in the Tube Gain circuit, stock resistors prevented the transistor from conducting enough to deliver enough plate voltage to the tube circuits, which was causing too much distortion,

a 15K/2W was substituted for R9- 56K/2W, which slams the base a little harder, increasing conductance and thus raising plate voltage,

a 15K/2W was substituted for R81-27K/1W which turns the transistor off more, thus dropping the plate supply more than before,

still, the max voltage is only 225 out of 280 applied to the transistor, however, if you use a 1K resistor for R9, the transistor conducts more voltage, but when the 1 meg pot is turned all the way off, R81 gets almost the full 280 V-dc on it, which means it has to handle 5 watts, this is no good on a 2W resistor, but with the two 15 K resistors splitting the 280 V-dc, power goes down to 1.3 watts each which reasonable.

eliminating the transistor circuit by bridging across the terminals with a piece of wire will deliver the full volts to the tubes for max headroom, but then you get hum, as you can see by the schematic, the transistor forms a pi network with the first cap and thus, pwr supply hum is reduced quite a bit, so that transistor serves two functions, pwr supply filter and voltage adjuster,

note that if a tube shorts, there are still plate resistors in series with the tubes so the transistor will not see a lot of collector current, and this TIP 50 is pretty strong, 400 V-dc max and 1 amp I-c, so the constant load of the tubes, being in the milliamp range, will never stress the transistor, if you have a constant load on a transistor circuit like this, then it becomes an adjustable voltage regulator of sorts, and also a current source,

the amp sounds nice now, does not overheat, makes the 15"  speaker cab dance across the carpet, and gets the 12 string guitar sounding like a sitar drone from 6 feet away so we bolted the lid on,   

 

[CJ]

here is the final chassis,

that toroid pwr transformer has options for pri hookup of 100 , 230 and 240, 

so in the US, you end up with +/- 80 V-dc rails on the MOSFET's instead of the schematic specs of +/- 65 V-dc, apparently, some transformer companies in Taiwan do not know that we run 120 over here,
in fact, i do not know of any country that runs 100 V-ac,
these filter caps are 5600uf ea rated at at...............80 V-dc! 

so we took 20 turns off each high current sec (100T>80T) that feeds the fets, now we have less stress on the filter caps and fets that run in a safer region with less heat, so a melt down due to a sensitive bias pot is less likely to happen, so we did not replace the stock fets as they run cool now at max power,

we also combined sec leads inside the xfmr so we have two extra power lugs for the B+ xfmr that was added to replace the funky HV wind on the toroid, this aux xfmr can be seen in the top right, the new 15K 2W carbon comps are seen near the top left,

 

[CJ]

well, looky here, you can avoid re-winding the pwr trans for 120, all you need is a 120 hookup diagram which is not shown on the schemo,

since this thing will run on 240, it will also run on 120 if we have two primaries, which we do,

just wire the toroid like the diagram in the red box, doh on me,  ???

 

[PRR]

> stock resistors prevented the transistor from conducting enough

The tube-stuff may act a lot like a 25K-100K resistor. (The first stage will act a lot like 200K, 2nd stage like 300K, WCF unsure but 12AU7 can't suck *that* much).

Say 50K.

With 280V available, the max current can be is <6mA.

With the transistor, we want the base resistance to be "small" compared to 50K * hFe.

Lo-volt work I would assume hFE>50. But as voltage rating goes up, hFE *must* be reduced (see Andrew Groove 1967). And TIP50 is quite an old design/spec, maybe not the best-levered BJT in the drawer. Also as you say it can go up to 1 Amp, so 6mA is far below what it is optimized for.

In fact it may not cut-off better than 1mA. Still plenty of room to our 6mA, but we sure are on the low end of its design range.

Some spec-squint suggests hFE could be as low as 10 at 6mA and 10V.

For most interesting conditions, the base resistance is mostly the 56K, || with other stuff, say 50K.

Then the emitter impedance will be near 50K/10 or 5K.

5K is small compared to our assumed 50K tube-load.

So it should pull-up to 254V of the 280V supply.

I would not expect a 254V/280V difference to be a big problem.

I wonder if Q1 could be shorted? Then your drop is all about R9 against ~~50K, and voltage sure will be low. And you were happier with R9 at 15K, which would carry the tube-load without any help from Q1.

There's ways a 400V part could die in a "280V" circuit.

However if the cab is dancing, I guess everybody is happy.
___________________________________

Parts of Japan ran 100V wall-juice; maybe still do.

This is an issue when US GIs bring their Japanese apartments back to the US. Some stuff suffers quietly and other stuff burns up.

80V on nominal 56V rails is Bad Idea.

As you discovered, if it has 100V and 240V, good odds it has 120V, just forgot to tell you. Being a mostly-US brand, it probably was wired 120v stock, they thought they only had to tell you the Export options.

 

[TheJames]

So what was your take on the SVT 3 from a sonic perspective?

I always liked them myself.  I'd rather use a SVT 3 than a SVT 4.  Full tube SVT's are nice, but I don't want to lug them around.

 

[CJ]

MOSFET's sound good for bass, i suspect they were trying to copy the Hartke 3500/5500 product line, which also has a graphic eq and compressor,

there is a cool 6 way tone switch on the SVT3 which uses a real inductor,

and the offset on the speaker terms zeners back and forth in the +/- 3 mv range, so the DC servo works great, even when i  mis-wound the prw trans the first time so it had 56 and 60 volts of AC, the servo balanced the rails to +/- 80,
this thing can deliver good bass without the output cap, seems to have subwoofer properties with the tone sw on max bass,

so bias the pwr fets out of cutoff, change vendors on the toroid pwr trans  and you have a pretty good amp,

180 turns of thick wire on the pwr xfmr sec means a pretty stiff power supply, which is probably why the cap dances on the carpet, this room is now earthquake prrof, as all loose tools are now on the floor,