What's a good target for input sensitivity for a 10W amp?

[rongon]

I have to build a stereo push-pull amp this summer, for my brother. I had an idea to do a PP 6AV5GA triode amp into 6.6k ohm 20W OPTs. If I use a Williamson-style topology with 6N6P tubes (high perveance, happy at low plate volts), there will still be a lot of gain open loop. I figure the input sensitivity will be something like 0.25V pk signal input to full power (I figure about 8 watts per ch). I could add 6dB of gNFB, but I don't think I'll want to add more, because he wants a tubey sounding amp to go with his vintage JBL speakers. (I think they're 4311, but I'm not sure.)

I could set him up with a 'passive pre' type of thing for volume control and switching between sources.

He has an RCA-style 12AX7 RIAA preamp with 12AT7 cathode follower output. The RIAA preamp has pretty hefty gain, something like 45dB.

His CD player is standard, probably puts out 2V rms signal for 0dBFS.

Would an 'integrated' amp that got to max output with only 250mV (or 500mV) pk input be way too sensitive, pick up every little noise, etc?

I usually aim for 1.4V pk to full output, but the Williamson topology has a lot of gain stages.

Anybody have an opinion on this?
--

 

[PRR]

Or you could build-in an input pad with a socket for a resistor to select on test.

 

[Wavebourn]

90 dB/W/M is good enough for 10W amp, if you mean speakers.
If to mean input sensitivity of an amplifier, it does not depend on output power. It depends on your music sources. For most modern sources 0.775V is a good standard.

 

[rongon]

Thanks.

For most modern sources 0.775V is a good standard.

If that's 0.775V rms, then that's about 1V peak.

But, 0.775V input for what output? To max power/clipping of amp output waveform?

 

[Wavebourn]

For a rated power how you define it. It is either power on THD less than stated, or a power that the amp can produce infinitely long.

 

[grovergardner]

I usually aim for 1.4V pk to full output, but the Williamson topology has a lot of gain stages.

Anybody have an opinion on this?
--

A Williamson with feedback will have plenty of gain for passive input from a CD player. I can't vouch for the phono stage. 🙂 I've built a very nice "original" Williamson using 807's and a pair of Heyboer's Peerless S-265-Q copies, which are about $450 a pair. It's lovely. I used the Williamson circuit documented here, with a few minor changes:

https://www.google.com/url?sa=t&sou...FjAAegQIBRAB&usg=AOvVaw0N0UjhSKXavv27D6Vo-Bvv

 

[kodabmx]

As other have stated, standard is roughly 1V pk.

If you're building the whole system, you can do what you want. On my power amps, full power takes about 6V drive. My line stage can provide 11V. I do that because I live in a high EMI/RFI zone, and the less gain in the power stage, the more stable the system is.

 

[grovergardner]

I would add that, in my experience, a Williamson-style circuit *without* feedback would be too sensitive for standard sources.

 

[Wavebourn]

I would add that, in my experience, a Williamson-style circuit *without* feedback would be too sensitive for standard sources.

Without feedback a pentode class AB amps has no meaning, you would not want to listen for such a sound.

 

[grovergardner]

Without feedback a pentode class AB amps has no meaning, you would not want to listen for such a sound.

Yes, for sure, but I think the OP was talking about building a triode amp.

 

[jan.didden]

The input sensitivity is not directly related to output power but to amplifier gain and max output level.

Say you have an amp with a gain of 20 and 10V (all RMS) max output, you need 0.5V for max output.

10W in 8 ohms is about 9V. So what's the gain of your amp?

Jan

 

[rongon]

The input sensitivity is not directly related to output power but to amplifier gain and max output level.

Say you have an amp with a gain of 20 and 10V (all RMS) max output, you need 0.5V for max output.

10W in 8 ohms is about 9V. So what's the gain of your amp?

Jan

9V peak = 6.363V rms = 10W into 4 ohms (my speakers are 4 ohms)

9V peak output divided by 0.3V peak input = 30x or 29.5dB

I have one of those little TPA3116 class D 20 watt per ch amps that has gain of about 27dB. I find it to be too sensitive, but it's usable.

9V peak output / 0.5V peak input = 18x gain, or 25dB

That would probably be better.

Let me re-phrase the question. What is a good target for a power amp's gain, in dB?
--

 

[rongon]

OK, so let's say I have a 50W (rms) amp.

Into a 4 ohm load, that's 20V peak output (14.14V rms).

The gain for 1V pk input to 20V pk output is of course 20x, or 26dB.

But...

If my amp makes only 10W rms into a 4 ohm load, that's only 9V pk out.
If 1V pk input = 9V pk output = 10W into 4 ohms, that's gain of 9x, or 19dB.

So...

If the target sensitivity for input volts to full power output is 1.1V pk (approx 775mV rms) then should low power amps have less gain?

See, that doesn't make sense to me. High power amps should probably have less gain, so pops, clicks, hums, buzzes and other noises don't blow things up.

I think...

 

[GoatGuy]

What is a good target for a power amp's gain, in dB?
—

Let's reduce it to a straight-forward equation.

V_{IN RMS} = something between 0.2 V and 1. 5 V, typically
P_OUT = a desired output effective power
Z_SPKR = nominal speaker impedance (which varies over 3:1 range)​

OK…

V_{OUT RMS} = √( P_OUT • Z_SPKR )
A_OVERALL = V_{OUT RMS} / V_{IN RMS} … reducing to
A_OVERALL = √( P_OUT • Z_SPKR ) / V_{IN RMS} … reducing to​

That settles that part. However, we're also talking about this in the vacuum tube sub-forum, so it is not quite right. We also need to include the output step-down ratio as well.

X = √( Z_PRIMARY / Z_SECONDARY )​

So, internal amplification factor is

A_INTERNAL = A_OVERALL • X​

All in?

A_INTERNAL = √( Z_PRIMARY / Z_SECONDARY ) • √( P_OUT• Z_SPKR ) / V_{IN RMS}
A_INTERNAL = √( Z_PRIMARY • P_OUT• Z_SPKR / ( Z_SECONDARY • V_{IN RMS}² ) );​

Note that Z_SECONDARY is almost always the same as Z_SPKR, so it further simplifies to

A_INTERNAL = √( Z_PRIMARY • P_OUT/ V_{IN RMS}² ); ​

Let's try it with SPKR and SECONDARY impedance being same …

Z_PRIMARY = 3200 Ω
P_OUT = 10 watts
V_{IN RMS} = 0.2 volts

A_INTERNAL = √( 3200 × 10 ÷ 0.2² )
A_INTERNAL = √( 800,000 )
A_INTERNAL = 894
A_INTERNAL = 20 log₁₀( 894 )
A_INTERNAL = +59 dB​

Just saying,
GoatGuy ✓

 

[Wavebourn]

If the target sensitivity for input volts to full power output is 1.1V pk (approx 775mV rms) then should low power amps have less gain?

See, that doesn't make sense to me. High power amps should probably have less gain, so pops, clicks, hums, buzzes and other noises don't blow things up.

I think...

More power is for less sensitive speakers or bigger rooms, so think about SPL in the listening position per input voltage, it should be the same no matter what is the output power of the amp. The sensitivity has to be equal to get 0 dB input voltage level.

 

[GoatGuy]

PS: you can see … why the input is usually specified at 0.773 volts RMS instead of 0.2 V_RMS as in my example. Puts a LOT less “strain on the gain”. Still … the output does need to produce (as in the example)

0.2 V_{IN RMS} × 894 = 178 V_{RMS INTO PRIMARY} effectively.​

Which… if it is stepped down from √( 3200 ÷ 8 ) = 20 times, is then

178 V_RMS ÷ 20 = 8.94 V_SECONDARY.​

If the speakers (as a sanity check) are 8 Ω… then

P = E² / Z = 8.94² ÷ 8
P = 10 watts​

Great!!! the formulary works!!!
Yippee…

GoatGuy

 

[DPH]

Take a step back, rongon. I think you're mixing a bunch of things together.

Let's start with some principles:
1.) input sensitivity is typically rated as the minimum input voltage required to achieve max output from the amplifier. One needs to set the specifications on what max output is (clipping vs a certain distortion level vs ...).

2.) Excepting compression, V_out = gain*V_in. Therefore, a small amplifier needs much much LOWER gain to achieve max V_out for a given V_in.

3.) Maximum SNR preservation and minimal external EM incursion (with a nod to stability) says use the absolute max output from the the earliest stage possible and minimum gain necessary later (the "gain early and gain once" principle, as I like to call it). This means find out what the max "good" output of your phono/cdp/sound card/whatever, and the minimum gain in your output amplifier.

4.) With diy projects, we have the luxury of measuring our input sources and design accordingly! So go measure what that is and set your output amplifier gain to be *barely* above the minimum necessary to satisfy Vout_max = gain*Vin_max.

5.) As GoatGuy also factors in, the voltage gain of the tube amp needs to account for the step-down "negative gain" of the output transformer. The gain I refer to above is from input terminal to output terminal, agnostic to internal gain structure.

Edit: haha! Looks like a bunch of us were in flight at the same time and I got distracted with other things in the middle of my post.

 

[45]

See, that doesn't make sense to me. High power amps should probably have less gain, so pops, clicks, hums, buzzes and other noises don't blow things up.

I think...

I disagree. My Roksan Caspian M2 power amp has 31.6 db gain and just over 85W/125W power into 8R/4R. No pops, clicks, hum, buzzes and other noises of any type. I drive it with any preamp with gain up to 12 db without problems.

If the amp is done right there is no arm in a little bit more gain.

Going back to your initial post, if you are worried about strange things happening then go for the triode PP with 6 db fbk. That will be a safe bet....

 

[GoatGuy]

(1) input sensitivity [needs to match sources]

(2) V_OUT = gain • V_IN. Therefore…
(2b) lower output amps need lower gain for same sources

(3) SNR and minimizing EM incursion … uses the absolute max gain from the earliest stage and minimum gain necessary later.

(4) With DIY projects … go measure sources and set your output amplifier gain to be barely above the minimum necessary to satisfy V_{OUT MAX} = gain • V_{IN MAX}i.

(5) GoatGuy noted that internal voltage gain of a tube amp needs to account for the step-down of the output transformer. The gain I refer to is from input to output terminal, agnostic to structure.

Thanks for the Kudos.

№ 1, № 2 are spot on.
№ 4 is likewise.

My only difference is with № 3, in that I personally don't completely maximize the first stage gain using the three main maximizers... LED-under-cathode or capacitor-bypassed-resistor or actual fixed negative bias, but rather the naked-cathode-resistor approach. I find it substantially linearizes the first stage valve, while cutting primary gain by ½ more or less. This is easily made up without adding audible floor noise in the 2nd stage, which also — for signal-contouring symmetry — should likewise have the naked-resistor cathode bias setup.

But that's just me.
Just saying,
GoatGuy ✓

 

[ALL EARS AUDIO]

If you have a scope check the input voltage just before clipping at 1khz if not .775v will do.