Vox T60 SS head low output

[Jack W.]

Here's the fault. Powers up, but only puts out approx 11 watts max before positive peak starts to clip. All caps have been replaced & resistors checked. New insulators & sweat solder joints. Check pwr xsistors on simple scope curve tracer ckt, trace showed no leakage.
Here's my test results so far:
Checked driver TX : DCR
Pri = 34.1
Sec1&2 = 4.4
Put 1Vac 1Khz in pri = .37 / .33 Vac on sec.
Insulation = >500M between windings & core.
Scope waves where clean.
Ckt readings: No signal = Rails (w/this sub PT) +/- 26Vac
Q4 Vdc
B= -.96
C= -12.5
E= -.79

Q5 Vdc
B= -.79
C= -9.6
E= -.63

Q6 across .5ohm E resistor = -5.7mV , B-E = -85.3mV
Q7across .5ohm E resistor = -7.1mV
Vdc before R22(180ohm) = -25V after = -12V
1khz thru input jack, volume full: @ Q5 collector = 10Vpp
start of pos clip. Output 8 ohm load = 27Vpp @ pos clipping.
AC specs at PT primary = Iac 410mA @ clip.

Winding short in driver TX?

[J M Fahey]

No, no winding shorts.
Remember that these transformer driven amps were very tightly specified.
Low output transistor beta + no transistor drivers + current-limited (by definition) Class A driver all conspire for the output stage *barely* having needed current drive.
In a way, it worked as a sort of very crude current limiting, maybe not as much as a "protection circuit"but as part of its signature sound.
I see your Q5 driver has 12V on its collector, instead`of specified 16.5V , also it clips on positive peaks.
Both symptoms match.
Just guessing, try to change its bias (increase R18 *up to* 100K) so its current lowers and you get those 16.5V back, and retest.
Did you change Q6? Because you might have a low beta device there.
Good luck.

[R.G.]

You can test the driver transformer for an internal short. Test circuit here: Transformer Short tester

It's probably not a shorted turn inside the transformer. If it was continuously shorted, the output would be about nil all the time, not just on one polarity of signal.

Q5 Vdc
B= -.79
C= -9.6
E= -.63

I actually think this is the problem in some fashion. The schemo wants the emitter to be up at about 0.9V - 1.0V. That voltage, reflecting the current through the 10 R emitter resistor, is the resting current of the driver. What is important about that is that the driver can turn on one of the outputs by pulling down and increasing the current through the transformer, but for the other polarity driver, it relies on decreasing the current already stored in the core of the primary inductance. The driver can only turn off by as much as it is on at idle. If that happens to be the drive for the positive side, the output transistor runs out of current and clips.

I'm a little surprised that the voltage after R22 is so low - 12V instead of 16.5.

But the less-than-expected current in the driver would account for one polarity of output not being able to move the output.

[Jack W.]

Thanks for the quick reply guys,
Someone else worked on this before I got it. The power xformer has been replaced & can't really verify it's the correct one for this amp. It's giving me the +/- 26Vdc rails. Seems low for a 50 watt amp.
Q5,Q6,Q7 are all NTE 104 replacements. Using a down & dirty hfe test on my bench DDM I get Q5=1116, Q6=923, Q7=954.
Anyway raised R18 on Q4 to 58k, it put 15Vdc after R22.
W/signal into preamp Volume full I get only 32.5 Vpp @ 8 ohm load.
Both Pos & Neg peaks clip uniformly. That's only 16.5 watts @ full power!

Could the low rails from this sub PTX be giving me such low power output?

[J M Fahey]

Oh well.
I think our early diagnosis was reasonably close.
Let's go step by step:
1) Those rails don't sound that bad to me, considering the germanium transistors.
*Maybe* the original one was +/- 30 V ... we are in the ballpark.
What *is* poor is getting only 32Vpp (altghough germaniums were noted for that).
Somewhere there is a very good article on old VOX SS amps (is it RG's? Don't want to be unfair) including all you need to know plus changing output devices to Silicon ; a smart move because you can get excellent easy to find transistors over the counter, and real ones, not the crappy (and expensive) NTE fakes.
2) Those incredible Hfe/beta numbers you seem to be getting, in fact come from the leaks, do not indicate actual gain.
I think it will pay to read and use said article.
3) Heresy/Plan B : I think the above plan is the best; now if you don't need a restoration but only the *looks* of that head , maybe you can slap an LM3886 chipamp there, which will work happily with those rails, and provide excellent (although not original) sound.
Before you stone me: I know (and work with) a lot of "Beatle Bands" which *kill* for stage looks (that's what their public pays for,anyway), and take great trouble to fill the stage with "VOX" amps, use Höfner violin basses, Rickenbacker and Gretsch guitars, Ludwig drums, etc. ; beyond the customary haircuts, clothing, and even the required pound of rings on "Ringo"'s fingers.
They are happy with authentic looking but reliable working amps.
They fill stadiums in Buenos Aires, plus one of our most popular Clubs: "The Cavern". (wonder why they called it that way )
They make huge contests where bands from neighbouring Countries get in fierce competition for a recording deal and a Tour to Japan.(Why to Japan? Don't know .... maybe the Japanese are also nostalgia lovers, even if ersatz)
Unbelievable?
Just check Google
and
The Cavern

Oh well.
At least I got to sell a couple dozen Vox lookalikes along the years.

[Jack W.]

Yes, I've read that great technical Vox article, it has help me with several Beatles amps in the past and come to think of it they use pretty low rails for how much power they produce. This amp has a buyer if it's in "original" working condition...these guys are a bit finicky with the "Beatles" sacred amps around here...and of course it's very easy for costs to exceed these relics. I'd like to avoid inverting the PS to use NPN's if possible.
So the question now is other than a bias change what else in this odd circuit would need changed to put in say silicon 2n2955?

[J M Fahey]

You needn't invert the PSU, Silicon PNPs are plentiful and cheap, but you'll need to rebias.
Refer to that article which spells it clearly and/or wait for ?RG? to mannifest for further instructions.

[Jack W.]

I found this quote from a thread in this forum under -Help with 1965 Vox solid state amp, please!- from R.G. on 07-04-2011 "You can convert to silicon PNPs, but it's harder with this version of the stacked circuit than with the Thomas Vox ones." discussing the T60 output. The one here is a JMI version.

Harder maybe because of the current limiting issues & the odd output stack?

If this amp was mine I wouldn't hesitate to experiment, damaging the driver TX is not an option. The output stack isn't setup like the Thomas Vox but the procedure described in R.G.'s articles should be about the same. (& using about six DDM's!)

I guess I'm looking for a proven solution for this odd Ckt.
So going with some MJ15016 PNP's for Q6 & Q7,
We'll start from there (& hope it doesn't turn into a table top cigarette lighter..)

Thanks for getting me this far guys..I appreciate it for sure.

[J M Fahey]

Harder because the biasing divider is made out of a 1K resistor (no problem here) and the DC resistance of the corresponding driver transformer secondary .... which you can't touch.
I have a faint idea floating, but first will need the DC resistance of the driver secondary plus the bias voltage developed across it.
Good luck.
PS: I don't care too much about those MJ15016, since they are just glorified higher tension MJ2955 ... which you don't need here.
I'd rather choose a transistor which at, say, 5A has the highest reaqsonable beta.
Maybe somebody can suggest one.

[R.G.]

Sorry I've been MIA on this one. I'll try to catch up.

First, there probably isn't any proven solution. Someone's done it, I'm sure; I've just never seen it and I look.

Second: what power can you expect?

If you are driving a resistive load (you're not, but this is how everyone calculates this), then you will get the power supply voltage minus any saturation and driving losses as a peak voltage on the speaker. If you have 26V, and you can saturate the output transistors to 2V each, then you'll get 24V peak on the output load resistor. From that you can calculate power.

P = V²rms/R = (24/1.414)²/8 = 36Wrms. If the load is 4 ohms, you get 72Wrms. Note that this only happens if you have a rock solid 26Vdc that does not sag and a rock solid 2V and no more loss in the output transistors. Neither of those assumptions apply in the real world, but there is some justification for calling it 30W into 8 ohms and 60W into 4 ohms. However, the schemo is inconsistent in saying or implying 60W at 8 ohms and 30W at 16 ohms. It does not have the power supply voltage to do that, at least not with sine-wave RMS power. 16 ohms power would be about 15W. Peak currents (we'll need those later) will be about 3A for 8 ohms, 6A for 4 ohms into a resistive load.

The OC28 listed on the schemo says it will support a 60VBVceo and have a gain of 15-30 at six amps of collector current and 1V collector to emitter. That means that it is unlikely that the power supply was much over +/-26Vdc, because the output devices would be prone to breakover from the voltage. Back when this was designed, no one really understood forward biased safe operating area, and it was thought to be OK to go to the edge of the BVceo. We now know that you have to make the transistor have a much higher BVceo to get to a safe operating area big enough to use in an amp. I would not put OC28s back in there even if I could get them, for this reason. However, at least the OC28 has a spec for gain at 6A, which is good. It wasn't a great transistor for that job, just all they could get at the time.

I would use MJ15025's. They cost $4.60 each from Digikey right now. They are an amount of massive overkill that pleases me. For the additional $1.60 each, you get 250V, 16A and 250W power dissipation in the same package, and the same gain. Once you get them running, they will live through things that would kill a dozen OC28s.

You should be able to swing the output to within maybe 2-3V of the power supply at peak. The OC 28 had a gain of worst case 15 at 6A, so the bases had to have 6/15 = 400ma to avoid running out of base current with a worst-case transistor. The windings on the driver show a 3:1 ratio. so it needs 133ma on the primary transistor to swing that. The driver setup as shown in the schematic does not have that; it has 0.9V/10 = 90ma through the emitter resistor of the driver, and that is a hard limit on one of the transistor base current swings. So: unless you have output transistors with gains over 6A/0.09A = 22, it will run out of base current driving a 4A load.

That may be OK. Or they may have hand selected for high-gain OC28s in the factory. The "typical" curves for the MJ15025 show gain at 35 for currents of 6A. It'll probably work.

So if you swap in $9.50 worth of PNP silicon devices, you need to bias them. This can go two or three ways. JM is correct, the resistance of the winding is being used as part of a voltage divider to provide the base bias. The bias needs to be increased from about 85mV where it is now up to about 400-450mV. You could just lower the 1K until the voltage rises to five times where it is now. That current is about (26Vdc -85mV)/1K = 26ma now. It would have to come up to about 130ma.

That is a bad idea. It puts 25 times the idle power in the driver winding.

You could put resistance in series with the winding to keep about the same current, but raise the DC voltage on base-emitter. It needs to be (450mV-85mV)/0.026A = 14 ohms. That increases the voltage the winding has to produce to get the current into the bases of the transistors. It increases by 14*133 = 1.86V on the secondaries and 5.59V on the primary. The driver may or may not have enough voltage available to do that.

You could convert to the USA Vox style biasing, with 1K to a new lower resistor divider, and the driver secondary between the resistor junction and the transistor base. That is probably a much better idea on both counts,

I'd use a resistor which gave a 450mV (about) voltage when fed from 1K and 26V; this is about 17.6 ohms; call it 18.

You could use the MJ1302A power device; it's not in a real metal TO-3, but it has a gain of about 80-100 at 5-6A. This *could* get you into trouble with oscillation because of the very much improved bandwidth. But I'd go with the MJ15025 in the metal TO-3.

Third: If you don't have an incandescent light bulb limiter, get or make one. It's how to prevent further damage after you change the power parts.

[Jack W.]

I was concerned about any adverse effect bandwidth change may introduce, hoping it would be manageable. Being so use to voltage concerns with tube circuits, R.G. clearly brings back the importance of "current thinking" with discrete xsistors. I think a bit of ckt modification similar to a USA Vox should give usable results. I have done a successful Buckingham conversion in the past. But as with most designs, the variables can be distracting. And yes putting in an IC chip will always be a cost effective, reliable option.

JM how many of those Beatle bands want their equipment original but use a wireless or pedal board? LOL!

Thanks guys!

[loudthud]

I have Jim Elyea's book Vox Amplifiers The JMI Years. Met him at the Dallas Guitar show a few years back and bought the deluxe edition with a book of schematics. The earliest T60 schematic shows 1-1.5 ohm resistors in series with the transformer secondaries with the 1K resistors connected directly to the base. In later versions these were deleted.

I've been wondering about the absence of diodes from Collector to Emitter on the output devices. These normally clamp any inductive spikes from the speaker that could kill the output transistors. Come to think of it, I don't remember seeing them on any of the Vox-Thomas models either.

[J M Fahey]

Don't wonder , they lacked many other things we take for granted.
After all, *they* were the pioneers in this.
They also lacked thermal tracking, Zobel networks, series inductors, short protection, you name it.
Anyway I admire them for what they did, of course.