voltage divider

[shooter]

I need to drop my signal volts from 70vpp to around 30.  I was wondering which, if either, is better, splitting the plate resistor or making the divider after the coupling cap feeding the next stage grid?

the plate R is 150K and the grid has a 1meg "gain" pot.  (creating 10% or so of compression would be fine)

thanks

[kagliostro]

Give a look here, may be there is something of interesting (or not)

http://el34world.com/Forum/index.php?topic=16371.msg160572#msg160572

K

[HotBluePlates]

I need to drop my signal volts from 70vpp to around 30.  I was wondering which, if either, is better, ...

Gotta see the circuit in question. There is no "better" method without knowing a specific application (because there are lots of good ways to reduce signal level).

- If you have 2x too much signal and have a volume control before-hand, turn it down. Reduction from "10" to "5" on the volume control reduces signal level to 1/10th the full-up volume.
- If you have a cathode bypass cap in an earlier stage, remove it.
- Design less gain in the preceding circuitry.

Etc.

Every resistor adds some amount of noise; every gain stage amplifies noise along with the desired signal. Having excessive gain which you have to knock down (outside of certain distortion applications) just worsens signal-to-noise ratio. Which is then another argument for designing from the output stage backwards to the input, to avoid building in excess gain and noise to be reduced late in the amp.

[shooter]

thanks guys n K, it was interesting.  HBP, I couldn't get enough drive with the NFB so I pulled it but now I have to much.  with NFB 12vpp, without 70.  I will probably revive the thread but was trying to keep my questions to a minimum n work thru this till I'm really stuck.
thanks again

[HotBluePlates]

HBP, I couldn't get enough drive with the NFB so I pulled it but now I have to much.  with NFB 12vpp, without 70.

Then use less feedback than the 12v case, but more than zero feedback.

[jjasilli]

Ditto.  Also, you can wire a pot as a variable resistor in the NFB circuit.  Dial it to where you like it.  Then measure the resistance of the pot at that setting, and hard wire a fixed resistor of that value in place of the pot.  This diagnostic use of pots is from Doug's Library.

[shooter]

I been messing with that, original  22k in series with 3.3uF non-polarized, I doubled the 22k n got less than a volt ac-pp, after that out to 100k with sub-volt change.  I was having guitar guys eval each change, they liked no NFB best but started getting minor squeal at vol=8, which makes sense since I'm hiting PSE el-84s with 70vacPP, so that's where I'm at.  I'm gator-clipping split load plate resistors in about an hour, setting all gain/vol n tone for max test signal at el grid, then playing with R values till I get about 20vPeak?  the els are cathode biased near the 250, 250 7.6 point into a 2.5k load n I remember HBP saying you may need up to twice bias volts for drive?
(I'm using 880Hz @ 500mV as input signal)

[HotBluePlates]

... I remember HBP saying you may need up to twice bias volts for drive? ...

All bets are off without a schematic.

My original comment probably described a push-pull output stage, where the bias voltage is known, and the output stage is Class A and driven just to the point of cutoff (just short of this is the likely limit of clean output power). When measuring "peak-to-peak" driving voltage, as an old tube data sheet would spec it, one meter lead is attached to the grid of an output tube on one side of the OT, while the other meter lead is attached to the tube grid on the opposite side of the OT.

Doing all that, the signal voltage to one side is driving the output tube to a positive peak equal to that tube's bias voltage (or the grid would momentarily read 0v with respect to the cathode), while the tube on the other side of the OT is being driven to it's negative peak (or the grid momentarily reads 2x bias voltage with respect to the cathode). So the "positive peak" plus the "negative peak" (because they're in opposite directions) is 2x the bias voltage.

the els are cathode biased near the 250, 250 7.6 point into a 2.5k load

7.6v of bias and you have a 70v peak-to-peak input? Or signal of 5x the bias voltage to each tube? That doesn't seem workable...

(I'm using ... 500mV as input signal)

Show me a guitar pickup that has 500mV of output... 100mV seems more likely to an input jack (measure a guitar's output & see for yourself).

So maybe the measurement technique or the test signal are incorrect, giving strange numbers to work with.

[shooter]

I probably am running my test signal HOT, but I'm trying to guess what a guitar with OD, fuzz, etc "could" put out.  Hopefully by spring i'll have some baseline numbers so I have some idea what to use.  And ur right 70vpp was not good, did work thu, had this grungy, highly distorted sound, but the squeal was bad enough the players backed off to 10 gain, 4 vol.
Anyway, I settled on 2 47k tapped in the middle then fed to the CF n on to the ELs.  Gives me about 33vpp with about 10% clipping in the + lobe that can be "dialed out" with the tone pots even while the gain n vol are dimed out. I didn't scope the ELs out, but will in the morn.   I attached the schematic, I'm using the bottom preamp (v3-4)  The top is just doodling - for now.  I did boost my preamp B+ so I might not have the 10% clipping now, will test in the morning.

[terminalgs]

I probably am running my test signal HOT, but I'm trying to guess what a guitar with OD, fuzz, etc "could" put out.  Hopefully by spring i'll have some baseline numbers so I have some idea what to use.

Vp-p expectations of a guitar signal: it depends a lot on pick attack. with hot pickups (say Z=10K-11K humbuckers) and hard attack,   you could see initial spikes around 1Vp-p, sustained 250-500mVp-p ??.  of course, lower for lower output pickups and softer attack.


fuzz doesn't necessarily increase overall Vp-p.  On a scope, a pedal like a univox superfuzz will show a lower total Vp-p, but it may appear to sound louder due to multiple audio waves added, all at different frequencies, some maybe square, etc.


for any effect, see the manufacturers sensitivity and gain specs (if/when they provide them).  a TC Elect boost pedal boasts +26dB of gain (conveniently, they don't use "dBV" or dBu" in that spec...).  put that in front of your amp, crank it up,  and you might see 43Vp-p spikes at the grid of V1!


You might design the amp to sound good with guitar only, no gain pedals. you can't predict what someone will put in front of it any differently than Fender, Marshall, Fischer, or anybody else can or did  (or attempted to..).   Also, you don't want special switches on it that require some engineering knowledge of the circuit (like "never flip with switch with both gain and volume knobs @10)..  If your amp sounds great, someone will want to borrow it, play with it, record with it, etc,,  and will flip switches without special instructions.


what does V4 buy you in this circuit?

[HotBluePlates]

a TC Elect boost pedal boasts +26dB of gain (conveniently, they don't use "dBV" or dBu" in that spec...).  put that in front of your amp, crank it up,  and you might see 43Vp-p spikes at the grid of V1!

They don't need to add the "dBv" or "dBu" because there's no reference level; instead, they're telling you the maximum amount of amplification the pedal can give.

For voltage (assuming equal impedances, which we'll ignore), dB = 20 log (Output/Input). Output over input is same as saying "voltage gain." Converting +26dB to amplification, we get:

26dB = 20 log (Output/Input)
26dB / 20 = 1.3 = log (Output/Input)   (To solve, take the "anti-log" of both sides, by raising 10 to the 1.3 power)
10^1.3 = 19.95 = Output/Input

So call it "voltage amplification of 20x."

With a 9v battery, it's physically impossible to get more than 9v p-p, and probably appreciably less. So 9v / 20 = 0.45v peak to peak input assures square-wave output. In reality, some lower input level will cause distortion, because the circuit has to have collector and/or emitter load resistances, and so probably can't throw an output signal as big as its power rails.

(EDIT: Added some notes to make the math process clearer)

[terminalgs]

thanks HBP, that's a good explanation. I hadn't considered the 9VDC power source.

[shooter]

Ya, I'm not designing for "what-could" be used, just trying to get a basic standard for me.  I wired it up, plugged in a cheaply strat n she sounded very good, ranged from nice bluesy funk to 90's grunge.  It's good for now, i'll get a full set of scope images and voltage readings next.  When I get the means I'm probably gonna change from PSE to cathode bias PP, but I still have 2 tubes to play with!  Thanks for all the help.

[shooter]

Sorry HBP, was posting while you were asking, then I saved as bmp, here's the preamp section
EDIT and PA section

[terminalgs]

what does V4 buy you in this circuit?

[DummyLoad]

what does V4 buy you in this circuit?

a white CF: with a Z out of about 40 ohms. why?

--pete


 

[terminalgs]

a white CF: with a Z out of about 40 ohms. why?

For two EL84s, it seems unnecessary, and would only contribute complexity.  if he was driving three 6550s or EL34s,  I could see  wanting a CF to drive the grids (like an SVT or something).


so, what I'm asking is what is the contribution  to the audio performance of the amp that V4 provides?


If it's just to experiment, I get it.  But if i'm missing something about the needs a twin el84s in parallel, I'm curious...

[shooter]

The amp was originally a proto build for a guy that wanted specific traits, to speed up the build I just bought a PCB preamp.  Now I'm using that board as my proto-type "base", every mod I do I pull out a component, add  wires n use gators to clip in values till I get smoke or pleasing sound!!  I'm working on my version of a 2channel "generic"  pre that will "mate" to your flavor PA, kinda a 2 board modular approach, well 3 with PS. So this 1st go-round is a basic fender-esk, pre, the other channel I'm still gazing at schematics, sometimes I think my wife would rather see me gazing at porn than schematics!

[shooter]

Need some help with the numbers.  I attached an XL sheet with scoped images n voltage readings.  When I went to figure amp power using Vsqr/R…..

((21.2*21.2)*.707)=318/R,  Using 8ohm, I get 39.75W, way high, but I’m using the 16ohm tap so I used 16 and got 19.9w which is about what I get figuring plate dissipation.  The problem I’m having is the 50% theoretical max for SE, which should put the amps power in the 9-10W range.  The 21.2 is VAC-PP at speaker, the .707 to convert PP to rms.  I used a 10ohm 250W R for load, is that my error?  the spkr dcr is 8.1, the R is 9.7.  I even took the 16ohm*  the 16% increase for the R vs spkr R n still came in at 17.34W

[sluckey]

the .707 to convert PP to rms.

That's not right.     V_RMS = .707(V_peak)

V_peak = .5(V_pp)

Your RMS numbers are twice as big as they should be.

[shooter]

Thank you, that gets me way closer to expected, still guessing R  thu, the tap obviously effects tube power, but what do I use for the math the physical 8ohms(or 10) or the ""tapped" value?
thanks again.

[sluckey]

You would use the value of the resistor, ie, 10Ω in this case. The power formula doesn't care about the OT. All it needs is the voltage measured across the resistor. If you measure 10V across a 10Ω resistor, that resistor is dissipating 10W. (I chose these numbers because I can do it in my head.)   

     P = E²/R = 10²/10 = 10

[shooter]

Thanks again, I wasn't sure about "effective resistance" vs real, I used real and the amp comes in at 5.6Wrms which now seems a little low!  21.2pp becomes 10.6p * .707= 7.5rms sqr ( 56.25rms)/10ohms.

I think i'll rate it as  11wPeak :)

At least my spreadsheet should have the right formula now!

[shooter]

I finished up channel 2 of my “grunges” amp.  It was basically a cut n paste from a hiwatt custom 20.  The poor EL’s were NOT happy with that much gain, so I swapped in a AU and it stopped V1 from breaking into oscillations.  I was also following MikeJ’s thread on white noise hum, n  I caught it!!, there should be a warning posted in the read 1st section!!

Using the AU fixed most n getting rid of the antenna farm I created on the PCB will probably fix the rest!

The tone is great, doing a cheesy mix with ch1 gets me everything from Death metal to twangish.

When I was scoping ch2, I noticed V1b had 100vdc at the plate, but the AC signal at clip was 110ACpp. (with AX tube).  Is that an indication I’m drawing grid current?

Also with my cheesy mix, ch2 comes in at about 90VACpp, ch1 about 30VACpp, out the best I can dial is about 23, is that a loading effect?  I was expecting 2 in phase signals to “add” mostly.

Thanks.

[terminalgs]

Also with my cheesy mix, ch2 comes in at about 90VACpp, ch1 about 30VACpp, out the best I can dial is about 23, is that a loading effect?  I was expecting 2 in phase signals to “add” mostly.

Thanks.

If the signals are 100% in-phase at point of mixing, they end up parallel with the same axis relative to ground, they don't add to one another.


also, is the power amp section still the same?  is the next thing the center lug is connected to the two parallel 470K grid leak  resistors? so that's 235K to ground.   if you don't have a coupling cap between the mix pot and the those grid leaks, between the mix, with an evenly centered mix pot at 50/50 mid point, I think you will attenuate the signal on either of the mix inputs to 21% (instead of 50%). 


If you want to up it, change the 1M mix pot to a 500K, or remove one of the 470K grid-leaks.


which is ch1? and ch2? ch1 is on bottom?

[shooter]

The PA section is still the same, CH1 is the tweed style, 2 the hiwatt.  There is about a 10% phase shift signal to signal, ch1 leads by that 10%.  I did "rethink" the questions, I was thinking when 2 different signals are mixed you get the sum, dif, n both fundaments, which has NOTHING to do with amplitude!!  so I did ask a dumb one there.  Also I found some info on grid current and to much drive, the article was talking PA section but I suspect it applies to the pre tubes also, if you overdrive by too much the tube will cut-off?  I also "remember?" at least in PA section there can be a 2X ht swing for the signal?.
I do have plenty of drive, actually too much, but I plan to change out the EL84's for AB EL34's which will probably need what I'm now attenuating.
Thanks for making the time
dave