screen and plate voltage

[Randall]

I just finished building a DR. Happy to say it looks like I got everything right, and it's quiet as a mouse. However, my screen voltages are sitting 2 volts above my plates, is this cause for concern? I thought they were supposed to be a few volts below plates? All other V's look good.

[Leo_Gnardo]

Lots of old amps have that "problem" yet function just fine. Many times I'll see them loaded with venerable, probably original installs, RCA & Sylvania 6V6's. I wouldn't worry about it.

[Chuck H]

That's probably "normal"? Every DR schematic (new and old) shows the plate and screen voltages as the same. I interpret this to mean they didn't want to show that the screens were a couple of volts higher, so they fudged the schematic Actually, the reissue shows the plates at 391VDC and doesn't show the screens, but does show the screen supply node (behind the g2 resistors) at 394VDC.

I'll bet if you cool the bias a tad the voltages will mate up or you might get the plates a couple volts higher than the screens.

[Randall]

Then I won't worry about it.

Another question about these amps is when I was putting it through the paces I found the Normal channel puts out 19 watta, but the Reverb channel only puts out 14 watts. I will scope the preamps to see if I'm losing something somewhere. The last one I built put out 19 watts as well. They are supposedly 22 watts, but are they really? I realize you can't tell the difference between 19 and 22 watts, I wonder if anyone else has seen this.

[jmaf]

How are you measuring power? If your multimeter is not true RMS the reverberated waveform may be putting the reading off.

Screen voltage will be above or below plate about 50% of your playing. Every time the plate swings the screen will remain in place and it'll be tens of volts above or below, no issues.

[Chuck H]

Is this a complete DR with the roach trem and shared cathodes, etc? Are there ANY!!!! differences from the AB763 the schematic at all?

[J M Fahey]

Another question about these amps is when I was putting it through the paces I found the Normal channel puts out 19 watta, but the Reverb channel only puts out 14 watts.

How come?
Power amp will clip at the exact same point no matter what.
Feed 100mV 440Hz or 1kHz into the preamp, set tone controls to 5, reverb and tremolo to 0 where available, and start slowly rising volume until power stage (scope speaker out) just starts clipping, then repeat through other channel.
They should clip at practically same level.
Just consider power clipping, what "number" is shown on the volume knob is irrelevant, if you worry because "channel 1 gives me 19W when on 7 , second channel on 7 just gives me 14W" that only shows a *gain* difference, which is normal and expected, but not the relevant point.
Just in case, repeat test with tone controls on 7 because those are not true Log pots so you are above the "jump".

I will scope the preamps to see if I'm losing something somewhere.

Gain may be different, but to drive power amp into clipping you need somewhat more than 1V RMS, *both* preamps are capable of putting out WAY more than that (think 20 or 30 V RMS) so they are not a power limiting factor.

The last one I built put out 19 watts as well. They are supposedly 22 watts, but are they really? I realize you can't tell the difference between 19 and 22 watts, I wonder if anyone else has seen this.

All the time; published power ratings go from "optimistic" to "absurd" .

From "100W" Fender and Marshall putting out 85/90W or barely reaching 100W but with visibly rounded peaks to "20/25W" 2 x EL84 amps actually putting out 14/15W to .... you name it.

In fact and to make the list shorter, only amps I saw meeting and exceeding "official ratings" are Ampeg V4/V22 etc. and American Marshalls ... but notice both:
* did not use 6L6/EL34 but way beefier Improved/Industrial 7027 or 6550
* used higher than normal 500/525/560V supplies.

Some 6L6 Fender DID meet and exceed 100W RMS .... pity they were sold as "135W"
and again: used scary high voltages *and* relied on Improved 6L6 , the Sylvania 6L6STR , not your run of the mill 6L6 .

In a nutshell: you can´t avoid Physics Laws; if you want high power *first* the PSU must be able to supply it and second the tubes must be able to handle it.

[Chuck H]

Power amp will clip at the exact same point no matter what.
Feed 100mV 440Hz or 1kHz into the preamp, set tone controls to 5, reverb and tremolo to 0 where available, and start slowly rising volume until power stage (scope speaker out) just starts clipping, then repeat through other channel.
They should clip at practically same level.
Just consider power clipping, what "number" is shown on the volume knob is irrelevant, if you worry because "channel 1 gives me 19W when on 7 , second channel on 7 just gives me 14W" that only shows a *gain* difference, which is normal and expected, but not the relevant point.
Just in case, repeat test with tone controls on 7 because those are not true Log pots so you are above the "jump".

Gain may be different, but to drive power amp into clipping you need somewhat more than 1V RMS, *both* preamps are capable of putting out WAY more than that (think 20 or 30 V RMS) so they are not a power limiting factor.

And this^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Juan went a little long (for better or worse ) But I hope you can see his point. If the channels were tested with the same volume knob settings then the results are arbitrary. You have to fix the input from the signal generator and the tone controls on the channels (and null the trem and reverb of course), then use the volume controls on the channels independently to determine the maximum clean output for each channel.

If there IS less clean power from the reverb channel under these testing conditions then the preamp is clipping ahead of the PI (which it shouldn't). This would indicate a mistake in the wiring or a flaw in the configuration you've used (which recalls my post #6).

[pdf64]

Another possibility is oscillation.
But I suspect Juan has identified what’s going on.

[Randall]

This is an exact full on repro of a AB763. Here is my methodology. Sine wave split into both channels, scope on dummy load. Reverb and trem dialed to zero. All tone controls on 5, repeated on 10. With Normal channel volume full on (read p-p, do the math) shows 19 watts. Turn down, repeat with Reverb Channel, different results, smaller p-p signal, math results in 14 watts.

Next I probe at the junction of the two 220K mixing resistors and the .001cap before the grid of the PI. Similar results, normal channel signal is strong and clean, reverb channel is smaller and a little irregular at full volume.

Comparing the second triode of each channel one at a time, the reverb channel actually is a tad stronger at this point. Where I am now flummoxed is I find the same strong signal (160v p-p) at the bottom junction of the 3.3M and 10pF mixing circuit after the verb channel second plate, but only 6 v p-p on the top side that feeds the grid of the second reverb triode. The 220k and 470K down by the reverb control appear to be good. Is it normal for the signal to drop so drastically here?

http://ampwares.com/schematics/deluxe_reverb_ab763.pdf

[Chuck H]

I have a guess about your results. I assume by your description that you're maxing the volume (of one channel OR the other) and then increasing signal from the signal generator until you see flattening on the wave forms. Is this right?

That's not the right way to do it with a guitar amp much of the time. I know it's the standard method for a lot of audio gear though. And it should have worked for the DR design. Many guitar amps have gain structures that could skew results. Still, try doing it like Juan described. That is, 100mV input, tone stack all fives, reverb and trem nulled. Then increase the volume knob ON THE AMP until you see flattening of the waveform. A very important aspect of this method is to MEASURE THE INPUT SIGNAL WITH THE UNIT UNDER TEST! You need to know that there is no signal loading skewing the parameters.

[J M Fahey]

Exactly, you are doing it wrong.
Your method optimizes setup for the higher *gain* channel, you inject "just enough" signal for it, and the lesser gain one is compromised.
And nowhere I saw you even mention clipping, which is the real Power Amp maximum power indicator.
You must inject "excess" fixed signal at the input, I mentioned 100mV; 200mV would even be better and any Tube guitar preamp will handle it, and *then* use volume controls to *reach* clipping ... which you did not.

Besides that, scopes are excellent to *show* waveforms and such, but are not the best *measurement¨* devices for the very good reason that human eye sharpness is part of the equation (fwiw *mine* is a mess ) so I MUCH prefer using a meter and reading "a number" on a display rather than counting grid divisions on a screen, any day of the week.

And scopes show you peak to peak values, meters show RMS values ... the traditional conversion factors (2.82:1) apply only to PURE sinewaves, not the case here by a Country mile.

Repeat test as suggested above and using a good RMS meter ... you will be surprised

[Randall]

Okay, now I'm really confused. With 100mv p-p 1K wave fed to both channels, tones on 5, trem and reverb zero, I see on the load 16 v p-p (16/2) x .707 = 5.6 vrms (4 watts). From the normal channel I see 9 volts p-p, just about 1 watt.

As suggested test again with tones on 7, Normal Channel = 17.5 v p-p, Reverb channel = 10 v p-p. No clipping at any point.

It is only when I drive the inputs with higher signal, near 400 mv p-p do I get higher output and clipping. What gives?

PS monitoring the input signal while under test shows no change.

edit: I just read the above post and repeated test using 200 mV p-p and my results are Reverb channel 12 watts no clipping, normal side 5 watts no clipping. Slightly higher readings from my Fluke 77 IV, but I don't think it is a true RMS meter.

[jmaf]

from my Fluke 77 IV, but I don't think it is a true RMS meter.

You either need a scope that shows RMS average or a multimeter that does. Whatever it's measuring now cannot be translated to RMS power.

[J M Fahey]

Okay, now I'm really confused. With 100mv p-p 1K wave fed to both channels, tones on 5, trem and reverb zero, I see on the load 16 v p-p (16/2) x .707 = 5.6 vrms (4 watts). From the normal channel I see 9 volts p-p, just about 1 watt.

As suggested test again with tones on 7, Normal Channel = 17.5 v p-p, Reverb channel = 10 v p-p. No clipping at any point.

It is only when I drive the inputs with higher signal, near 400 mv p-p do I get higher output and clipping. What gives?

PS monitoring the input signal while under test shows no change.

edit: I just read the above post and repeated test using 200 mV p-p and my results are Reverb channel 12 watts no clipping, normal side 5 watts no clipping. Slightly higher readings from my Fluke 77 IV, but I don't think it is a true RMS meter.

When I say 100 or 200mV I mean RMS, not peak to peak.
You are using 1/3 of tha signal needed there.


And in any case, signal must be high enough for amp to **clip** ; if not , rise it until it does.

*ALL* tests you did so far had insufficient signal level to drive amp into clipping, so none is valid to determine power output.
At some point I also suggested setting tone controls to 7 instead of 5, because real world pots are not *really* log.

But the main point is: no clipping? ... you have NOT reached maximum power.

Absolute worst case, IF your amp has less gain than normal, feel free to rise input signal whatever is needed until it reaches clipping, and measure there.

Most normal amps reach clipping with 100 or 200mV RMS at the (preamp) input, but if yours has low sensitivity, feed it whatever it needs.

EDIT:

It is only when I drive the inputs with higher signal, near 400 mv p-p do I get higher output and clipping. What gives?

Confirmation that you are using a way too weak test signal:
* 100mV RMS=300mVpp
* 200mV RMS=600mVpp
* 400mV RMS=1200mVpp

repeat tests actually reaching visible clipping and post results.

In any case, your amps seems to be having somewhat lower gain/sensitivity than normal/standard, but we´ll address that later.