Amp should be 100 watts but low on power, why?

[soundmasterg]

I decided it would be best if I start a new thread rather than hijack someone else's. I'm adding in the previous posts from the other one so people can see what was being talked about. I appreciate any help that anyone can provide.
Greg



It started life as a Sovtek MIG 100U, and I kept the power transformer because it had the correct voltages and was running a quad of EL34's, so I figured it should keep up. I modified it into a clone of a VOX AC100, with a second channel with higher gain in addition to the stock Vox single channel. I'm using the same phase inverter as VOX used, a 12AU7 seesaw variant of the paraphase. To mix the channels, I added another triode (half a 12AU7) in parallel with the first triode in the phase inverter. They share a plate resistor and cathode resistor, and the second triode in the phase inverter has it's own plate and cathode resistors for better performance. The VOX circuit had both halves of the phase inverter sharing the cathode resistor.

I seem to have a problem in the phase inverter area as I only get around 23v AC signal out of the first half of the phase inverter, and this is with a 55v signal coming out of the high gain preamp. The bias is -35v in fixed bias and around 29v in cathode bias. I pulled the output tubes and get about the same signal swing without them in there, so I don't think the output stage is affecting things. I built the stock Vox circuit with the single triode and a Fender LTP in an outboard chassis and subbed it one at a time into my amp in place of my phase inverter circuit, and I get around the same performance...which is not enough. The supply voltage for the phase inverter is around 410v, so there should be plenty of signal swing. The signal looks good on the scope too. I've checked and verified all of my wiring several times, and checked parts values and everything appears correct. The amp is perfectly stable with low hum and good tone, no oscillations that I can hear or see on the scope....yet I still have low power.....does anyone have any ideas of what kind of things I should check or where I should go from here?

Thanks,

greg

POSTED by soundmasterg
--------------------------------------------------------------------------


What's your screen voltage and idle plate current?
__________________
Jon Wilder
Wilder Amplification


POSTED by Wilder Amplification
--------------------------------------------------------------------------


Hey Greg,

Check the power tube grid waveforms on a scope. Can the PI drive the grid voltage right up to 0V? If so, the power tubes will clip it with grid current there, so the PI is doing as much as it can be expected to.

If it doesn't make it to 0V then you need more PI output. With 35V bias, you need 70V p-p of drive for each grid, which an AC voltmeter would see as 25V RMS.

Now, what about the screen voltage and screen resistors? If the screens aren't getting enough juice, you can drive the control grids as hard as you like, but the output will still be lacking. KT88s need a bit more than EL34s.

Next, what about the power supply, does it sag excessively under load?

If you put EL34s back in, does the power output go up to what it should be?

Etc.


POSTED by Steve Conner
--------------------------------------------------------------------------

I'm not sure how to tell if the PI is driving the grid voltage up to 0v? I think I was getting about 20v AC out of the plate of the first half of the PI when I had a clean signal coming into it, and the second half of the PI was a couple volts less than that. Those numbers go up as I run the preamps up more of course, but then its distorting the signal past the RMS power level where I should be checking for output power of course.

The screen voltage is about 460v or so I think and the plate is at 470v. I have 1k 5w screen resistors, and I'm using KT77's, not KT88's. I haven't tried EL 34's in it yet, but have a set of used old stock Mullards that I can try. I doubt if those will make a huge difference however as a JJ KT77 is pretty similar output power to an EL34.

The power supply doesn't seem to sag a whole lot under load. I forget what it was and will have to measure it again tomorrow, but it wasn't going down lower than 400v on the screen when loaded if I remember right. I post what it is tomorrow. The power transformer was running a quad of EL34's at 485v, and three 12AX7's in the Sovtek. In my amp, I have an EF86 and 12AX7 in one channel, and a 12AU7 and 12AX7 in the second channel, and a 12AU7 seesaw paraphase inverter, and then the quad of KT77's. In place of lots of 12AX7's and 12AU7's, I've used a couple 12DW7's so my overall tube count is lower. With the extra tubes over the Sovtek, the filaments are still at 6.3v, so no issue there.....and the B+ drops to 470v in my circuit instead of the 485 in the Sovtek. I'll gret some numbers for you tomorrow and thanks for the help!

Greg


POSTED by soundmasterg
--------------------------------------------------------------------------



...assuming the power supply is providing sufficient Vp and Vs, here's the basic equation for tube/transformer PP output power:

Po = rp' * (Vg*gm)^2

...where:

Po = output power (at primary-side of OT)
rp' = effective plate load resistance (tube and reflected load)
Vg = control grid rms signal voltage
gm = tube average transconductance

...the only "tricky" part is rp', the effective plate load resistance combination of tube rp and reflected speaker load Zo back through OT.

...notice that the equation is simply: Power = I-squared x R, expressed in vacuum tube variables, ie: Ip = Vg*gm.
__________________
...and the Devil said: "...yes, but it's a DRY heat!"

POSTED by Old Tele Man

[soundmasterg]

Ok, so I messed around with the amp a bit today. Here are the measurements on the amp.

IDLE - FIXED BIAS

Plate = 466v
Screen = 450v

DIMED - FIXED BIAS

Plate = 422v
Screen = 371v

IDLE - CATHODE BIAS

Plate = 435v
Screen = 420v

DIMED - CATHODE BIAS

Plate = 423v
Screen = 373v

I measured output power and I get about 37.4 watts clean in fixed bias and 44.5 watts clean in cathode bias. That seems backwards to me! Dimed I get 61 watts fixed bias and 54.3 watts cathode bias. This is with a quad of JJ KT77's and a primary impedance on the output transformer of 1750 ohms. I tried a quad of OS Mullard EL34's and the power output was about the same. Each power tube is around 40ma of idle cathode current. I'm not sure how to measure the plate current at the moment.

I don't think I get enough drive out of the phase inverter. When measured clean in fixed bias, I get about 23v AC at the plate of the first triode in the phase inverter, and 17v out of the second triode. I know they aren't balanced all that well right now but I'm not worried about that as much as the low power. I don't think it gets back to 0v on the power tube grids. I looked at it on the scope, and I switched the scope from monitoring AC to DC and it was below the 0v line. Is that how you check?

Greg

P.S. OTM, that math just flies right over my head.

[Steve Conner]

You mentioned you were using a 12AU7 as the PI. Did you try a higher gain tube in there? I've tried the 12AU7 in a floating paraphase, but it didn't have enough gain to make a balanced output: one output was considerably bigger than the other.

I used a pot in place of the "see-saw" resistors to trim for equal outputs, but putting in a 12AT7, which was what I ended up using, would be easier.

How did you measure that screen voltage, before or after the screen resistors?

Is that how you check?

I set the scope input selector to "Gnd" and adjust the beam to the top of the screen, so the top of the screen is 0V. Then I set it back to DC, and look to see if the peaks make it to the top of the screen. Even if the PI is perfect you'll see them flatten when they get there, because of grid current in the power tubes, but they should make it there.

[soundmasterg]

You mentioned you were using a 12AU7 as the PI. Did you try a higher gain tube in there? I've tried the 12AU7 in a floating paraphase, but it didn't have enough gain to make a balanced output: one output was considerably bigger than the other.

I used a pot in place of the "see-saw" resistors to trim for equal outputs, but putting in a 12AT7, which was what I ended up using, would be easier.

How did you measure that screen voltage, before or after the screen resistors?



I set the scope input selector to "Gnd" and adjust the beam to the top of the screen, so the top of the screen is 0V. Then I set it back to DC, and look to see if the peaks make it to the top of the screen. Even if the PI is perfect you'll see them flatten when they get there, because of grid current in the power tubes, but they should make it there.

The Vox AC100 uses a 12AU7 stock and gets 105-115 watts of power according to Bruce @ Mission amps. So I should be getting that with this too....but yes I did try a 12AX7, which had a marginal amount of increased gain....like a couple volts, but nothing earth shattering. The balance between the two sides isn't that far off either....I could dial it in closer and plan to but haven't at this stage because the low power is a larger issue. I measured the screen voltage at the socket, after the screen grid resistors.

Ok, using your example of the PI waveform, I am not getting to 0v. I did look at gnd and it was set in the middle of the screen on my scope, and when I set it to AC and look at the waveform, it is at least one full graduation down at full power (dimed), and lower than that when I'm looking at RMS clean power.

I had thought it could be my mixing arrangement with the phase inverter so I tried the stock AC100 setup in a seperate chassis and jumpered it in to mine, while disconnecting the circuitry in mine for the phase inverter, and it was actually a couple volts lower in swing than my setup with the mixer. Pretty weird! I can't mess with it more until the weekend though.....working too much.

Greg

[soundmasterg]

Well just an update here.

I discovered what my low power problem was today thanks to a suggestion from PRR over at the Hoffman board. When the parallel triodes on the first half of the phase inverter are in operation in this amp, normally only one channel is in use. When this occurs, the unused triode brings the plate voltage down due to it's idle condition, and this stunts the gain out of the phase inverter. I built the stock AC100 seesaw paraphase inverter into an outboard chassis and tested it in place of my mixer phase inverter in my amp. Previously I had done this but not got any better results because I didn't disconnect all the parts in my amp that I should have for the phase inverter circuit to make sure they weren't interacting. When I disconnected everything I should have to do that as a proper test, I was getting more signal swing and higher gain out of the stock Vox phase inverter. So my mixer circuit has to go and I'll have to redesign the amp to mix with resistors instead of the extra triode. I think I will utilize the now unused 12AU7 stage on the input along with the stock 12AU7 stage and make a parallel triode. This will bump the gain up a bit to make up for the loss using the mixing resistors, and will keep the channels in phase still.

Thanks for all the help, especially to Enzo here and PRR over at the Hoffman bbs. Sorry Enzo for the unexpected PM.

Greg