Design considerations for high voltage, high power, tube output section/PSU

[loudthud]

I'm having some difficulty in imaging a situation where that would be true. Perhaps you could illustrate how this could be so with a schematic example?

I disagree with Kevin on some things, but on grounds, he is usually right on. I don't see a star ground in the pics, but they are usually not implemented well with a simple stack of spade lugs on a single screw. From the symptoms I would first check that the oscillation is not coming down the B+ line to the preamp. It could be just the ground on the preamp filter cap because that cap is a low impedance, and any noise on the ground goes right to the B+.

[nsubulysses]

Thanks for your suggestions KOC and LT

I will try grounding the input jack to chassis rather than connecting it to the other grounds.

LT, the "star" ground is right next to the bias supply on the right side of the turret board, held down with a shakeproof washer on the turret board standoff. There are three wires going to it, white wire from power supply grounds, a black wire from speaker jack and power tube cathode grounds, and the 3rd wire which is also black is from input jack and tone and volume control grounds. Perhaps those should be grounded to the cathode bypass caps of their respective stages?

LT, when you say the preamp filter cap is low impedance and could be coupling to B+ do you mean V4 or V3 capacitor ? would you recommend a resistor in series with ground between PS and preamp capacitors then, or something?

I'll mess with this more tonight and see what I can come up with. Any other suggestions please keep them coming and I appreciate it.

[catalin gramada]

Too many joints more than 2 wires called star. Keeping mind-each stage with own decoupling cap share a node and from this nodes each node with one wire to only one spot. This spot is the star ground.The wires from each node to star should assure just supply return and no signal return through it. This spot (star) should be tied to chassis.Unfortunately in rc coupled stage the resistor should be tied to the most sensitive side node so it will be a long ac path return for previous stage.some situation claim a bus ground as a better solution in term of noise.
If you have not choices due to weird layout choose to keep signal return shorter and supply return longer and not reverse
In some cases can consider the chassis as a big star spot.but the rest of rule is compulsory

[loudthud]

LT, when you say the preamp filter cap is low impedance and could be coupling to B+ do you mean V4 or V3 capacitor ? would you recommend a resistor in series with ground between PS and preamp capacitors then, or something?

The RC filters in the B+ line should act like Voltage dividers for noise. The resistor is the "series" element and the capacitor is the "shunt" element. For maximum attenuation of noise, the shunt element should be as low an impedance as possible, and it needs to go to a clean ground. If there is noise where you ground the capacitor, it gets coupled to the B+ almost without any attenuation. The resistor will isolate the noise upstream (where it will do less harm) if the capacitor is going to a quiet ground. Putting a resistor in series with the capacitor's ground will only let more noise through.

Kevin will be more than happy to tell you which of his books has the best description of how he does grounds. I respect Kevin's rights to his own intellectual property, so I won't scan and post his stuff here. If you go to my profile and look at what is posted under the "messages" tab, you will find links to threads when I have diagrammed how I would do grounds. This amp has lots of gain so grounding is critical.

The problem you described above where the control setting affects the frequency of the motorboating reminds me of one of my builds where there wasn't enough filtering in the B+ line, but a noisy ground can cause the same problem. With the OT near the input, you may need a shield between the OT wires and the input jack. You may also have to isolate the input jack ground form the chassis by some means.

In general, the ground of the input jack should go the the ground end of the cathode resistor of the first stage. The filer cap for that stage should also ground at the same place. Ground should follow the signal path stage by stage picking up filter grounds as it goes back to the phase inverter.

[nsubulysses]

Hallo. just at quick first sight. Take off the input jack and move it in the hole between gain pot and power switches. It is no ideal but just can give a suggestion if your problem gone. If is so further you can move away two holes to the right those switches in far empty holes and screen the input jack with a shield.

You guys it sounds pretty wonderful !!!

I can put volume and gain in 10 now with guitar or. NoI guitar plugged in and it will not sound the horn !!

[nsubulysses]

I was wrong to dismiss Catalin's idea about the input jack .Messing aroudn with the old alligator clip and aluminum foil makeshift shielding I found that the input jack did pick up significant noise from the OT. I guess I thought it wouldn't be an issue since it's a sheilded wire anyway, but it definitely was.

I think the main problem though, which is why it was mind boggling, was that I had a poor solder connection in the power supply (probably screen supply ground) that was causing some wild instability. I realized that when the amp was on I could chopstick around alittle bit and get a little crackle, so I resoldered all the PS connections. Flipping power back on with input jack relocated and PS connections re-flowed I could put Gain and Volume on 10 without oscillation!

Can't thank you all enough for your input. Now to just hone the sound and let it rip

[catalin gramada]

Enjoy !

[nsubulysses]

Hum at full volume is not objectionable with a 300 ohm 20W resistor between plate and screen nodes but chokes are cheaper than I thought and I think I would like to try one to see the difference.

I'm looking at Hammond 193 series. My B+ is about 575V so perhaps getting a choke with 600V max operating voltage is pushing it. Not sure.

Should I just get a a 800V rated one, 100mA or more that has the lowest DC resistance possible?

Seems like Hammond 193L and 193J could work good. Both are 800V rated
193L - 5H, 300mA, DC resistance 57 ohms
193J - 10H, 200mA, DC resistance 87 ohms

My B+ node is 220uF/700V and my screen node is 110uF/700V. I believe that ties into the equation right?

[KevinOConnor]

Hi Guys

The 600V choke is fine at 575V. Just like a filter cap, this is working voltage.

Have fun

[pdf64]

I emailed Hammond to enquire whether a TR choke would be ok in a higher voltage application, with the response

The 194B would be OK at the 580V B+ .

We 100 % test this model at 1000 Vac ( 1414 V peak ) for isolation.

Continuous at 580 VDC would be fine

[nsubulysses]

Would a 90mA choke be cutting it close though in heavy overdrive situations? I will basically be playing this full volume only.

[catalin gramada]

Hello. It is hard to say how will influence performances of your amp. Some guys will see just ripple rejection.but what consider more important is how it influence linearity and dynamics of your power supply with direct consequence about your amp presentation.Inductance and voltage drop by momentary current through .That can made a difference between a sterile and an amazing amp.It is difficult to predict.I don't know if there are recipes for that. think need to be very familiar with the circuit to do the perfect choice.The high cap values don't help to take easy conclusions. You need to experiment some how.
For the ripple there are on line calculators can use.
Just an advice- don't be tempt to use too small values which in conjunction with filter cap risk to create a resonant circuit into near audio range
Or just pick.up a convenient large value
. say 3.5- 7H 150 ma arround with a limited dc resistance at 100obm or so and be happy with that. I did aso many times and I.m happy man

[nsubulysses]

These JJ 6CA7s seems like they can't handle this high voltage. This is really unfortunate and what I was hoping would not happen, but I thought it would be ok. FWIW the JJ logos are not blackened at all. Still as bright red as the day they were new it seems. I am biasing in the 50-60% range.

I have 1.5K 5W screen grid resistors. B+ and screen supply are about 570 and 565V. This is only about 20-30V higher than Ampeg V4 B+ and screen and JJ 6L6GC works great in that amp. AND 6CA7 has higher voltage ratings on the datasheet for B+ and screen than 6L6GC. What the hell.

When I put them in at first the amp did 24.5V at clipping into 4 ohms. Then after a week I checked again and it did 23V. Then after practice last night I checked again and it's down to about 22V, so 121W at clipping when it started at 24.5V, 150W at clipping. I have about 15-20 hours on these power tubes only

Putting in new JJ 6L6GC the amp does 24.5V into 4 ohms again. Using a quad of old, poorly matched GE 6CA7 the amp does 24.5V again. I wonder if the old poorly matched GE 6CA7 would be able to maintain a high output power if I put some hours on them. That would be cool, but it doesn't help me because I need to use new production tubes in this amp.

Anybody have any thoughts or ideas on what I could do? I guess I could just try 6L6GC next but I really thought 6CA7 would be a better fit with it's 800V max plate voltage and 500V max screen. 6L6 is 500V max plate and 450V max screen. Again, what the hell. Works good in V4 at ~540V plate and screen!

[pdf64]

It may be worth checking whether it's just one of the 6CA7 that's gone weak (rather than all 4 rapidly wearing out).
eg compare the cathode currents of each at high signal level.

[nickb]

As Enzo has said, it's watts that kill tubes, not volts. With that in mind, how much are they dissipating at idle? Can you reduce that a bit and still get acceptable crossover distortion?