• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

first DIY amp start up-yowza HELP!

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Well, fired up the amp for the first time Suday night. With 5AR4s out, measured 6.7v for filament and 5v for rectifier filament. A little high on 6.3v filament but close to 5%-so far so good!

Next up, 5AR4s and 0A3s in, measured all B+ points on the left side and was within 5v-success!

Then as I started measuring B+ on the right side I was @ 20v low.


then all of a sudden the 0A3 on that side started sputtering and flashing so I shut it down.

Re-checked the circuit, looking for shorts, double-checking against the schematic and couldn't find anything wrong. Thought I may have a bad 0A3 so I swapped a new one in and tried again.

This time sputtering and flashing from both 0A3s, so I shut it down and went to work with my vom. Found the dropping resistor, 200ohm 10w from the rectifier to the 0A3 that initially flashed had opened up completely. I'm guessing that was the cause of the lower B+ on that sidebefore it opened up and shot the voltage up on the remaining 0A3 causing it to sputter and flash.

Now I don't know if I just had a bad dropping resistor or if something caused it to open. Both sides are identical, so if one side works correctly then I'm leaning towards a bad dropping resistor-but how can I find out for sure?

Could I have damaged the 0A3s or are they tough enough to have survived?

Here's a link to the schematic. The only deviation being using 100uf 350v filter caps off the rectifiers rather than 150uf 300v, all other filter caps are correct capacitance, but over-rated by 50-300v.


Power xfrmr used is custom wound by Heyboer and is 800vct@700ma, 5v@10amps and 6.3vct@10amps.

Speakers were hooked up to the output xfrmrs before the amp was turned on to check filament.

Any and all help would be apreciated!

mr mojo


Joined 2003
Paid Member
> measured all B+ points

Did you measure the -20V BIAS voltage???

> was within 5v-success!

Moot fluke. Tubes don't need precise voltage, and ~1% results is just coincidence.

> the 0A3 on that side started sputtering and flashing

A gas tube, especially one that has slept for decades, will flash like an old fluorescent lamp on startup. In this silly circuit, the gas-tube won't start to start until the 6AN8 and 7591 warm-up. So at first I wondered if this were "normal".

But then you burned the 200Ω 10W resistor. This took over 200 milliAmps!!!!

Take the 7591s out and be sure you have a large negative voltage at their grids. Negative: check the polarity!!! And check that fiddling R49 R50 changes the voltage. Leave it set as negative as it will go, like -30V or so. Certainly more negative than -21V.

Now replace the 200Ω 10W resistors and put in the 7591 at V3, the socket connected to J4. Put a voltmeter on J4. Watch as power comes up. This point MUST not go over 1VDC. If the negative bias is cranked maximum negative, the J4 voltage may be very small, like 0.1V. That's fine. If no smoke or light-show after a few minutes, twiddle R49 to change the bias and verify that you can raise J4 to 0.2V or 0.3V. Don't try for 0.9V, since we only have one side of the OPT flowing current.

What I suspect is: you had little, no, or reverse grid bias on the 7591s, cathode current shot from ~70mA per pair to >200mA per pair. This smoked the 200Ω 10W resistor. Meanwhile G2 current went from ~10mA to >50mA, over-exciting the gas tube. But the real flaw is mis-biased super-heated 7591s.

> don't the 0a3's need a minimum load before operating?.....else they'll have no 'brake' on their processes

Note that this is NOT the conventional voltage regulator scheme. In fact it is a DE-regulator. But mainly they are trying to waste 150V to stay inside 7591's screen rating.

If current is too low, the gas tube will not have the specified drop. That's not a problem because in normal use, once the vacuum tubes are working, 6AN8 and 7591's G2 draw more than the ~5mA minimum current. 6AN8 current is well fixed by resistors. 7591's G2 current is a function of its plate current and bias and drive. In normal use, it probably won't exceed ~30mA per pair, so the gas tube is not over-cooked. If the 7591 is mis-biased, both plate and G2 current soar. Not to infinity, but probably to more than the gas-tube is rated; however the 200Ω resistor "won" the race to burn-up.

> Could I have damaged the 0A3s

Don't know. I've managed to avoid grossly over-currenting a gas tube.

But don't care. Check the 7591 bias voltage, stick the 7591s in. If the gas-tube survived, you will have 370V-420V at 7591 G2 and you will be able to set 7591 cathode jack voltages to 0.9V. If the gas tube died, I would bet on finding zero volts at 7591 G2. I rate a shorted gas-tube very unlikely (though its 0.01 bypass cap could short); if that did happen you'd find ~480V at 7591 G2, which is certainly wrong (they are rated 400V-440V).

What noises were the speakers making while the amp was smoking?

I'd be willing to bet you think at this point that I'm making an awfully good case for the power of natural selection-heh, heh.

As always I appreciate the detailed and thoughtful response. Just try and imagine for a second-if you dare-where I would be without your help!

After replaying the series of events, I think this is what I did. I plugged in the 7591s and 6AN8s to check fialment voltage. Next, although it was plainly stated in the instructions and I knew to do so, I'm pretty sure I skipped over adjusting bias at this point.

I'm an idiot-it's OK, I can admit it.

Next I plugged in the 5AR4s and 0A3s and started measuring voltage.
Spent a few minutes turning the amp off and on to alligator clip leads for different voltage tests on the right side so the amp was on for maybe 45 seconds to a minute at a time. All the voltages on the right were correct.

Then I started doing the same thing for the left side and noticed voltages were @ 20v low. Didn't know what the deal was, so shut it down and checked against the schematic for errors, couldn't see any, so started it up again and that's when the VR on the left side popped once and I shut it down.

Came back the next night and this time adjusted bias for -18 for each channel. I don't remember what they were adjust for initially, but I do remember they weren't correct. Started the amp up again to check screen voltage on the left side and about the time it got to 350-400 volts there was a light show under the amp so I shut it down immediately. Since the amp was upside down I couldn't get a clear glimpse of what was creating the light show, but I THINK it was the VR tubes.

If I follow your reasoning, initially I most likely had the left side biased less negative than the left and was burning up the 200ohm dropping resistor while I was checking the right side voltages. By the time I got around to checking the left the resistor had gotten hot enough to change value and effect my voltages and create the current rush in the VR tube.

When I fired it up again, even with the bias set at -18v, I'm guessing that was the last straw for that dropping resistor and it burned out, causing a voltage and current spike on the other VR tube and dropping resistor.

Last night I replaced both dropping resistors and both bypass caps on the VR tubes since they were now leaking DC. I've also got what looks like 2 burned carbon film resistors on each of the DC balance pots that need to be replaced as well.

As to the speakers, there was an occasional quiet pop and sizzle that sounded a lot like caps breaking in-which may or may not have been the case.

Tonight, after replacing the resistors on the DC balance pots, I will be following your instructions.

But, as you may imagine, I've got a few questions.

Next, why do you refer to this as a silly circuit? I'm not getting my back hairs up about it, just interested in your opinion.:D
Circuit came from a 1961 issue of electronics world and was designed by the tube engineering division of Westinghouse.

Next, as to the way the VR tubes are used, I know it's backwards. They usually will be used to keep a steady 75v. Here they're dropping the 480v to 405v for the screens. Aren't they regulating voltage the same way just in reverse?

Thanks again for your support of the local village idiot, he certainly does appreciate it.



Joined 2003
Paid Member
> support of the local village idiot

We all tread the same road. Those of us who already passed a pothole should send word to people who haven't yet got there.

OR: why do you think I know about forgetting to set G1 bias first, before power-up? Been there, done that, still smell the smoke.

> Next, why do you refer to this as a silly circuit? ...was designed by the tube engineering division of Westinghouse.

Ha! If I worked for Westinghouse, I'd like it too.

If I worked for a transformer maker, or a rectifier maker, I might skip the gas tube and use winding taps to make ~400V for G2 and ~500V for the plates.

For myself: just feeding G2 and Plate the same, within their ratings, gives 71% of the power, 1.5dB less peak volume, fewer bottles and less heat. I think the lily is over-gilded.

> Aren't they regulating voltage the same way just in reverse?

Consider some (made-up) numbers. Raw B+ is 475V, gas-tube drop is 75V, G2 gets 475-75= 400V. Now say the raw supply sags 20% (either from load or because the utility company sags). Raw power drops to 380V. The gas tube still drops about 75V. The new G2 voltage is 380-75= 305V. 400/305 is a 24% drop. A 20% sag has turned into a 24% sag. It anti-regulates.

If you also design the power transformer carefully, you can get an amp that makes rated power in rated load, but sags badly enough under a shorted load to semi protect itself. This is commercially useful (fewer blow-ups under warranty).

> issue of electronics world

Oh??? Were they using the same draftsman as Popular Electronics in that period? My "silly" was partly directed at the ugly drawing (looks good on the page, but hard to read as a circuit), which was often true of PopTron.
PRR-you're OK in my book!

This project has been a lot to bite off for a first-time scratch build so I guess I just got a flashing, high-voltage representation of the learning curve last few nights.

Two years what I knew about audio equipment started and ended with the "on" button. Restored a few old Scotts and started thinking-well how hard can it really be to build one of these from scratch?

I think I've got my answer.

I think we share the same thoughts about the Westinghouse engineers-the article is geared toward drumming up interest in the then "new" 7591.

As for the drawing, I kind of thought of it as a "circuit/drawing for dummies" which, in my case, seems to fit rather well.

Point well taken on both the overly complicated and reasoning behind the ultimately poor solution to screen regulation. At least they're pretty when they're not doing a damn fine impression of an arc welder, so I guess I got that going for me.

With a power xfrmr rated at 800vct@750ma, will I ever have to worry about circuit load sag?

If I do manage to get this thing running and post some pics of the heap, I hope you'll do me the favor of weighing in on my execution of the circuit-I'm always open to constructive criticism and I'd appreciate your thoughts.

mr mojo

I had a thought about the anti-regulation of the 0A3s as used in this circuit.

True, they do not hold voltage at a steady rate as they would if used for 75v, but I think they are regulating as they are used here, but rather than strictly regulating voltage, they are regulating the ratio of voltage between plate and screen.

Take for example true screen regulation held at a steady 405v. If B+ is at 475v and for some strange reason were to dip to 400v, then the grid will conduct current rather than the plate.

The way the 0A3s are used here, they will always maintain the correct ratio of voltage, almost as if they're being used as a dynamic variable resistor.

Wouldn't this dynamic ratio be more important than strictly regulating the screen voltage?

Hope you don't think I'm picking a fight-I just enjoy learning through discussion and reasoned arguments.

I am only an egg.

mr mojo

I think we're on the same page here, if B+ should be at 480 and the 0A3 drops the voltage to 405, then we've got a 1:1.9 ratio.

If B+ goes to 500v and the 0A3 drops the voltage to 425, then we have a 1:1.8 ratio

If B+ drops to 450 and the 0A3 drops voltage to 375 then we have a 1:1.2 ratio.

Until I actually did some simple math I mistakenly assumed x-75 would always be a 1:1.9 ratio. I'm sure more math minded folks would've known that without doing any math, but I'm a graphic artist not an engineer!

As I said before, I can understand the idea of regulating a ratio of plate/screen voltage as being more benificial than strictly regulating the screen voltage without regard to plate voltage.

What I seem to have with the 0A3s is a regulation system that keeps this ratio close, but nothing near exact. Better than regulating screen without regard to plate, but is there a solution which would keep this ratio exact?

Still grokking.

mr mojo


Joined 2003
Paid Member
SY is of course right. And if you are graphic-brained, you can graph the voltages and see that G2 will fall to zero while Vp is still ~75V; or if Vp rose to 2,000V(!) then G2 would be 1,925V, "same-as" 2,000V for any practical purpose.

The misregulation is not a problem in normal operation. I'm being picky about inelegance, and excess bottles. But then, excess bottles is not always bad. And not everyone can enjoy gas-tube glow.

> for example true screen regulation held at a steady 405v. If B+ is at 475v and for some strange reason were to dip to 400v, then the grid will conduct current rather than the plate.

No. In Pentodes (including the Power Tetrodes), the G2 current is a small fraction of plate current for any decent plate voltage. "Indecent" plate voltage would be like under 50V: as the plate current curve drops to zero, the G2 current goes high. But this is also where you can't get much power from the plate. So we pick a load-line that hits near the knee of the plate curves, not over where they are near vertical. Then the G2 current is still small.

Exact ratio of P/G2 DC voltage is easily done with a resistor divider. To ensure no sag as G2 current changes, divider current must be much larger than G2 current. G2 current is much less, but not much-much-less, than plate current. So the simple resistor divider is eating as much current as the plates, which means total chassis heat is doubled. SY suggests a smaller divider with a buffer: I've done that, but it is an inelegant can of worms. Not so bad with a MOSFET buffer, but you still wonder if you really want a Ratio or a Fixed Voltage. Use a tube buffer, and it is a semi-power tube with a cathode voltage so high it needs its own heater winding.

If the transformer is low-sag, but utility power wanders, a tapped power transformer gives fixed ratio of Plate and G2 supply coltage.

But considering the wide range of speaker impedances, I don't see how any plan to hold G2 within 10% of "goal" is justified. "8Ω" can be 6Ω or 12Ω, and will be 50Ω at bass-bump; what matters some 1.7:1 or 1.8:1 "ratio error"? Keep G2 high enough to deliver plenty of load current, or as much as you can afford, first-cost or replacement-cost (guitar amps run high G2 voltage to improve power/weight, at the expense of more tube replacement, but a guitarist is presumably getting PAID and can afford the cost of doing business).

I can find a lot of Old Men commercial designs that used a G2 voltage divider, especially in radios where there is a large current in the RF/IF tubes to swamp the output's G2 current variation. Also a lot of amps where G2 is fed direct from plate supply, or through a small dropping resistor. This forces the plate to run not much higher than the G2 rating: this fact seems to have driven tube design.

For maximum power with easy drive, G2 should be fairly low while Plate voltage should be high. This leads to tubes like 6550 and 8417, where the max-power condition is roughly Vp=600V, G2=300V. Note that a simple ceter-tapped transformer, or a voltage-doubler, can produce solid 300V and 600V with low parts count.

Most of the 7591 designs I recall ran the plate a little but not a lot higher than the G2 rating. Probably around 425V for both, or 450V to the plate and a dropper resistor to give 425V-400V at G2.

> the Westinghouse .... article is geared toward drumming up interest in the then "new" 7591.

I remember.

And they cranked it up to the very biggest numbers you can get from a 7591 without breaking the warranty. What does it claim, 50 watts? A few years before, 36 watts from a pair of larger harder to drive 6L6 was "great", so 50W from small sensitive 7591 is amazing. But most real-world designers tended to get 35W-40W from a set of 7591. Gas tube sales are good for Westinghouse, but bad for profit. (I have seen, here, a PA amp which used the gas-tube dropper trick, and in that case I think it was so they could push the ratings to the very edge, yet have the amp punk-out benignly when abused.)

FWIW: the 7027 has a 100 watts/pair rating; nobody ever got more than 35W-70W out of them. The 100W rating requires regulated 800V: if it goes higher, the base burns up, any lower and you don't get 100W. G2 voltage is also dead-critical and must be regulated. Of course in tube days, a regulator was essentially an audio amp with a DC signal: the same parts used AS an audio amp would double the power. So 100W with 7027 is really done with four 7027 and a 500V-600V raw supply, not two 7027 and a regulated 800V.

The original 7027 100W rating might have made sense in somebody's AM radio transmitter. An 800V regulated supply might have been available; might have aided compliance with regulations; anyway the cost vanishes in 10,000 watt output stages and a 1,000 foot tower.

I really appreciate your time and detailed, thoughtful responses-I've learned a great deal from this thread.

It's forums like this and folks like yourself that make the internet such a great thing-an exchange such as this simply wouldn't have been possible 15 years ago.

All the screen regulation info you've given me will be put to use on the next amp-nobody can build just one!

I'm with you on this particular solution-it's not the most accurate, it's not the most elegant-but when they're not sputtering and arcing, they sure are the prettiest.

Thank you for walking me through all the reasons and implementations for and against regulated screens and regulated plate and screen ratios. This is the kind of little stuff that, built over time leads to a holistic understanding of exactly what's going on in a tube amp and why-and it's these little things, and their implications, which aren't covered in detail in the books I've been grinding my way through.

As for the amps output, the voltages on the circuit are given under no signal. Elsewhere in the article it states the 7591 is capable of 50wpc, but at the cost of higher distortion, less tube life and more elaborate power supply. Artilce has the plates at 450 and gives a 35wpc estimate. Pretty durned close to your approximations!

Today, with the help of our only local tube tech, I hope to bring it on-line with a variac, multiple meters and a signal generator. If this thing actually works I hope to post some pictures this weekend.

Thanks again PRR, I really do appreciate your time and help.

mr mojo
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.